# Table of Contents - [About Oracle Mode | Flowith Official Document](#about-oracle-mode-flowith-official-document) - [Flow Mode User Guide | Flowith Official Document](#flow-mode-user-guide-flowith-official-document) - [Introduction to Flow Mode | Flowith Official Document](#introduction-to-flow-mode-flowith-official-document) - [Welcome to flowith | Flowith Official Document](#welcome-to-flowith-flowith-official-document) - [Flowith Knowledge Base Use Cases | Flowith Official Document](#flowith-knowledge-base-use-cases-flowith-official-document) - [Fundamentals of Prompt Engineering | Flowith Official Document](#fundamentals-of-prompt-engineering-flowith-official-document) - [How To Build Your Knowledge Garden | Flowith Official Document](#how-to-build-your-knowledge-garden-flowith-official-document) - [Introduction to Flowith Knowledge Garden | Flowith Official Document](#introduction-to-flowith-knowledge-garden-flowith-official-document) - [Advanced Features — Node References | Flowith Official Document](#advanced-features-node-references-flowith-official-document) - [How To Improve Your Prompt | Flowith Official Document](#how-to-improve-your-prompt-flowith-official-document) - [Current Capabilities Supported by Neo | Flowith Official Document](#current-capabilities-supported-by-neo-flowith-official-document) - [How to Report Issues | Flowith Official Document](#how-to-report-issues-flowith-official-document) - [Information Gathering + Organization Maestro | Flowith Official Document](#information-gathering-organization-maestro-flowith-official-document) - [Agent Mode User Guide | Flowith Official Document](#agent-mode-user-guide-flowith-official-document) - [About Agent Neo | Flowith Official Document](#about-agent-neo-flowith-official-document) - [How To Monetize Your Knowledge Garden | Flowith Official Document](#how-to-monetize-your-knowledge-garden-flowith-official-document) - [Agent Neo Use Cases | Flowith Official Document](#agent-neo-use-cases-flowith-official-document) - [Prompt Engineering 基础 | Flowith Official Document](#prompt-engineering-flowith-official-document) - [Subscriptions and Billing | Flowith Official Document](#subscriptions-and-billing-flowith-official-document) - [欢迎使用 flowith | Flowith Official Document](#-flowith-flowith-official-document) - [流式聊天模式介绍 | Flowith Official Document](#-flowith-official-document) - [高级玩法 — 节点引用 | Flowith Official Document](#-flowith-official-document) - [知识花园功能简介 | Flowith Official Document](#-flowith-official-document) - [Oracle 适合解决什么样问题 | Flowith Official Document](#oracle-flowith-official-document) - [Oracle 目前所支持的能力 | Flowith Official Document](#oracle-flowith-official-document) - [知识花园案例 | Flowith Official Document](#-flowith-official-document) - [如何建立和使用知识库 | Flowith Official Document](#-flowith-official-document) - [关于 Oracle 模式 | Flowith Official Document](#-oracle-flowith-official-document) - [Oracle 模式高级玩法 | Flowith Official Document](#oracle-flowith-official-document) - [flowith ナレッジガーデンの概要 | Flowith Official Document](#flowith-flowith-official-document) - [ナレッジベースの構築と使用方法 | Flowith Official Document](#-flowith-official-document) - [关于订阅 | Flowith Official Document](#-flowith-official-document) - [如何反馈问题 | Flowith Official Document](#-flowith-official-document) - [Oracle モードの高度なテクニック | Flowith Official Document](#oracle-flowith-official-document) - [信息搜集 + 整理大师 | Flowith Official Document](#-flowith-official-document) - [Flowith へようこそ | Flowith Official Document](#flowith-flowith-official-document) - [Oracle モードはどのような問題解決に適していますか？ | Flowith Official Document](#oracle-flowith-official-document) - [flowith ナレッジベース・ユースケース | Flowith Official Document](#flowith-flowith-official-document) - [Oracle がサポートする現在の機能 | Flowith Official Document](#oracle-flowith-official-document) - [ご契約に関して | Flowith Official Document](#-flowith-official-document) - [高度な機能 — ノード参照 | Flowith Official Document](#-flowith-official-document) - [知识花园变现指南 | Flowith Official Document](#-flowith-official-document) - [Flow モードの紹介 | Flowith Official Document](#flow-flowith-official-document) - [情報収集＋整理のマエストロ | Flowith Official Document](#-flowith-official-document) - [問題の報告方法 | Flowith Official Document](#-flowith-official-document) - [Oracle 模式使用指南 | Flowith Official Document](#oracle-flowith-official-document) - [Oracle モードについて | Flowith Official Document](#oracle-flowith-official-document) - [flowith ナレッジガーデン収益化ガイド | Flowith Official Document](#flowith-flowith-official-document) - [Oracle モード ユーザーガイド | Flowith Official Document](#oracle-flowith-official-document) - [プロンプトエンジニアリングの基礎 | Flowith Official Document](#-flowith-official-document) - [流式聊天使用指南 | Flowith Official Document](#-flowith-official-document) - [Canvas Navigation in Agent Mode | Flowith Official Document](#canvas-navigation-in-agent-mode-flowith-official-document) - [Flow モード ユーザーガイド | Flowith Official Document](#flow-flowith-official-document) - [Current Capabilities Supported by Oracle | Flowith Official Document](#current-capabilities-supported-by-oracle-flowith-official-document) - [Email Protection | Cloudflare](#email-protection-cloudflare) - [Neo について | Flowith Official Document](#neo-flowith-official-document) --- # About Oracle Mode | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2FBOw5ufjnihpyHsRl6OPP%2F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png&width=1248&dpr=4&quality=100&sign=82f401d1&sv=2) [](https://doc.flowith.io/agent-mode-neo/about-oracle-mode#what-is-oracle) **What is Oracle?** --------------------------------------------------------------------------------------------------- Oracle is a cutting-edge intelligent agent system developed by the flowith team, designed with the goal of achieving pre-AGI (pre-Artificial General Intelligence). It is specifically engineered to handle multi-step, complex tasks that require the use of multiple tools. Key features of Oracle include: * Autonomous task planning * Task decomposition * Unlimited tool utilization * Self-optimization Oracle's external capabilities are continuously updated and upgraded, allowing it to become increasingly powerful over time. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2FFe0YxPmfFaclcndbPzUc%2FAmazing%2520Mockups.jpeg&width=768&dpr=4&quality=100&sign=e138a96b&sv=2) Oracle Recipe and Nodes In other words, Oracle is a **data-level** (operating at the data layer, without physical medium or real-world interaction) **generalized** (capable of performing general tasks, not limited to pre-trained or pre-defined functions) **Agent** (beyond just an LLM, able to utilize tools, manage memory, and collaborate with multiple LLMs) that can complete complex tas**ks** (beyond simple queries or commands) and adapt to new situations (capable of self-learning and processing in novel environments or tasks). [](https://doc.flowith.io/agent-mode-neo/about-oracle-mode#core-concepts) **Core Concepts** ------------------------------------------------------------------------------------------------ **Recipe Workflow** The term "recipe" originally refers to a set of instructions for preparing a dish. In the context of Oracle, a Recipe represents a systematic workflow that AI executes to complete user tasks. When a user inputs a requirement or instruction, Oracle breaks down the overall directive into smaller subtasks based on its capabilities and available tools. These subtasks are then passed to a task allocator for step-by-step execution. After designing a complete Recipe, Oracle always confirms it with the user before beginning execution. During this confirmation process, users can directly modify specific steps within the Recipe, including: * Adjusting the order of steps * Deleting steps * Modifying text * Adding new steps ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2FBOw5ufjnihpyHsRl6OPP%2F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png&width=768&dpr=4&quality=100&sign=82f401d1&sv=2) A Recipe Generated by Oracle **Dispatcher** The Dispatcher, also known as a Task Allocator, is a crucial component in the execution of subtasks generated by the Oracle. For each Oracle-generated subtask, a dedicated Agent within the Dispatcher is responsible for allocation and implementation. This Agent identifies the most suitable tool from the Oracle's supported toolset and then executes and completes the subtask. For example, when the subtask is "**search for user reviews of flowith on social media**", the task allocator will invoke the "**Twitter Search Tool**", "**Reddit Search Tool**", and "**Google Search Tool**" to conduct searches for Flowith user reviews. You can view the currently supported tools for Oracle on the Oracle conversation initiation page. This list is updated in real-time, and we continuously add the most useful tools to Oracle's supported toolkit. [Current Capabilities Supported by Oracle](https://doc.flowith.io/agent-mode-neo/current-capabilities-supported-by-oracle) ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2Ffef0NUW8zbPpdLgMeTJ5%2F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE4.jpg&width=768&dpr=4&quality=100&sign=f8e39647&sv=2) Tools supported by Oracle **Memory and Context Management** Oracle is designed for complex tasks, which often generate extensive context due to their multi-step nature. If we were to provide all previous context to the LLM (Large Language Model), it could potentially impact its performance. To address this, Oracle employs intelligent context management during task allocation and execution. For example, when Oracle reaches step 10 of a process, its memory manager evaluates which information from previous steps is essential and incorporates only the necessary context into the current step. This approach eliminates concerns about excessively long context. **Self-Iteration and Update Recipe** After completing each step, Oracle automatically updates the entire Recipe based on the execution results of each step. For instance, if it fails to find the required information, it will incorporate additional search steps into the subsequent stages to ensure successful task completion. When Oracle inquires about user satisfaction with the results, and the user expresses dissatisfaction, it will generate additional steps in an attempt to provide the desired outcome. Last updated 10 months ago Was this helpful? --- # Flow Mode User Guide | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) [](https://doc.flowith.io/flow-mode-general-mode/flow-mode-user-guide#getting-started) Getting Started ----------------------------------------------------------------------------------------------------------- After clicking "New Conversation": 1. Enter your question in the chat box. Ensure that Oracle mode is turned off. 2. If you need to customize the model, you can select it from the options in the lower-left corner. 3. Click "Start" to begin the conversation. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252F5E7r38dogWlg9sBeOY3H%252F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%25BC%2580%25E5%25A7%258B%25E5%2588%259B%25E5%25BB%25BA.gif%3Falt%3Dmedia%26token%3Dcfba0670-cf84-486a-9ca7-a722e3578e99&width=768&dpr=4&quality=100&sign=94b6429e&sv=2) Start a New Conversation [](https://doc.flowith.io/flow-mode-general-mode/flow-mode-user-guide#follow-up-questions-on-ai-responses) **Follow-up Questions on AI Responses** ------------------------------------------------------------------------------------------------------------------------------------------------------- 1. Click on the node you wish to inquire about further. 2. Enter your follow-up question in the dialog box. 3. Press the Enter key. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FKOXp0Y6nY2BjJ2adjSVI%252F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E7%25BB%25A7%25E7%25BB%25AD%25E8%25BF%25BD%25E9%2597%25AE.gif%3Falt%3Dmedia%26token%3D2cb696d3-e97c-4c88-b8f7-99735fa39c02&width=768&dpr=4&quality=100&sign=cad2c92&sv=2) Follow-up Questions for AI Responses [](https://doc.flowith.io/flow-mode-general-mode/flow-mode-user-guide#regenerate-content-in-specific-nodes) **Regenerate Content in Specific Nodes** --------------------------------------------------------------------------------------------------------------------------------------------------------- When you're not satisfied with the content in a node, simply click the "Rerun" button to instantly generate new content with a single click. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252Fp5WRaOEG0k8kSaiYUtEF%252F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E9%2587%258D%25E6%2596%25B0%25E7%2594%259F%25E6%2588%2590%25E8%258A%2582%25E7%2582%25B9%25E5%2586%2585%25E5%25AE%25B9.gif%3Falt%3Dmedia%26token%3D94420c67-14a2-4da0-a9de-67426414ef2d&width=768&dpr=4&quality=100&sign=61fd53f3&sv=2) Rerun Node with One Click [](https://doc.flowith.io/flow-mode-general-mode/flow-mode-user-guide#expand-node-contents) **Expand Node Contents** ------------------------------------------------------------------------------------------------------------------------- If a node contains too much information to display conveniently: 1. Select the node. 2. Click "Show More". ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FCeD6XMVRTyk1l4p4DZO5%252F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%25B1%2595%25E5%25BC%2580%25E8%258A%2582%25E7%2582%25B9%25E5%2586%2585%25E5%25AE%25B9.gif%3Falt%3Dmedia%26token%3D27047a83-9e75-4f94-8c24-ce6dda6f1e46&width=768&dpr=4&quality=100&sign=1670341f&sv=2) Expand Node to View Full Content [](https://doc.flowith.io/flow-mode-general-mode/flow-mode-user-guide#copy-results-from-a-node) **Copy Results from a Node** --------------------------------------------------------------------------------------------------------------------------------- 1. Select the node. 2. Click "Duplicate this flow". ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FcZrE8cDqoEUjQJg49n9n%252F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%25A4%258D%25E5%2588%25B6%25E7%25BB%2593%25E6%259E%259C.gif%3Falt%3Dmedia%26token%3Dfbceb455-3ab0-471f-a108-51e961a75bb2&width=768&dpr=4&quality=100&sign=7ceb1734&sv=2) One-Click Copy for Generated Content [](https://doc.flowith.io/flow-mode-general-mode/flow-mode-user-guide#deleting-nodes) **Deleting Nodes** ------------------------------------------------------------------------------------------------------------- **Deleting a Single Node** 1. Select the node. 2. Click "Delete". ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252Fk2hJ9PGrMxNYtDcT6hFe%252F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%2588%25A0%25E9%2599%25A4%25E5%258D%2595%25E4%25B8%25AA%25E8%258A%2582%25E7%2582%25B9.gif%3Falt%3Dmedia%26token%3Dae727355-ba76-488f-afcf-9371cf40d1c7&width=768&dpr=4&quality=100&sign=5356e51b&sv=2) Quick Node Deletion **Deleting Multiple Nodes** 1. Hold down the "Shift" key and click on each node you wish to delete. 2. Press the "Backspace" key. 3. Confirm the deletion. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FixkYUJ3RVpizDiLDU4be%252F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%2588%25A0%25E9%2599%25A4%25E5%25A4%259A%25E4%25B8%25AA%25E8%258A%2582%25E7%2582%25B9.gif%3Falt%3Dmedia%26token%3De7ecca10-fe92-4d42-a1e8-dab1869135d9&width=768&dpr=4&quality=100&sign=3b0f0968&sv=2) Press and Hold the "Shift" Key [](https://doc.flowith.io/flow-mode-general-mode/flow-mode-user-guide#sharing-your-flow-with-friends) **Sharing your flow with friends** --------------------------------------------------------------------------------------------------------------------------------------------- 1. Click on "Share" in the upper left corner of the page. 2. Set the permission to "Can View". 3. Copy the link and share it. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FgjRi3HWmQX3zgc3IZLs6%252F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%2588%2586%25E4%25BA%25AB.gif%3Falt%3Dmedia%26token%3D3ff90730-ec18-4ae7-a735-555089f12bcc&width=768&dpr=4&quality=100&sign=a9227edd&sv=2) Invite Friends to Appreciate Your Creativity [](https://doc.flowith.io/flow-mode-general-mode/flow-mode-user-guide#collaborate-with-friends) **Collaborate with Friends** --------------------------------------------------------------------------------------------------------------------------------- 1. Click on "Share" in the top-left corner of the page. 2. Set the permission to "Can Edit". 3. Copy the link and share it. 4. Use the comment feature for communication within the canvas. 1. Click on "Add Comment" in the top-left corner of the page. 2. Click on the node you want to comment on. 3. Click "Add Comment" again in the top-left corner to exit comment mode. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FIXz3WRpSEjNTZpIRR8h8%252F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E9%2582%2580%25E5%258F%258B%25E5%2585%25B1%25E5%2588%259B-%25E7%25BC%2596%25E8%25BE%2591.gif%3Falt%3Dmedia%26token%3D182bf38e-5221-45e8-91d1-878762ff126e&width=768&dpr=4&quality=100&sign=eb2eb4f9&sv=2) **Enable Editing and Share** ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FFtL0BL11v8n9p0MFuICQ%252F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E9%2582%2580%25E5%258F%258B%25E5%2585%25B1%25E5%2588%259B-%25E8%25AF%2584%25E8%25AE%25BA.gif%3Falt%3Dmedia%26token%3Da53727ec-db2e-4a72-a911-af1f3ca93b31&width=768&dpr=4&quality=100&sign=b9a0bdb4&sv=2) Click on Specific Nodes to Add Targeted Comments [PreviousIntroduction to Flow Mode](https://doc.flowith.io/flow-mode-general-mode/introduction-to-flow-mode) [NextAdvanced Features — Node References](https://doc.flowith.io/flow-mode-general-mode/advanced-features-node-references) Last updated 10 months ago Was this helpful? --- # Introduction to Flow Mode | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2FUuJYJ2e6PcTCbp1xaI9z%2F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE2.png&width=1248&dpr=4&quality=100&sign=6e1ea374&sv=2) Flow mode is the next-gen AI interaction method designed specifically for content creators. It serves as the standard interaction method in flowith. Flowith utilizes an infinitely large canvas, enabling you to simultaneously engage with multiple AI assistants and generate the content you need. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252Fml5Et2WjMKKXe8wiZuhl%252F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE2.png%3Falt%3Dmedia%26token%3D409e6823-4a80-4746-a610-b689c53ebb4e&width=768&dpr=4&quality=100&sign=d2061b44&sv=2) **Why use Flow Mode？** If you only require AI for simple translations or basic question-answering, a conventional chat mode might suffice. However, if you need to engage in multi-turn conversations with AI, modify previous inputs, or collaborate with multiple AI assistants simultaneously, Flow mode offers a completely different creative experience. **Advantages of Flow Mode：** 1. **Multiple AI instances can generate content simultaneously in parallel threads.** 2. **You can repeatedly modify the AI prompt and generate prompts multiple times.** 3. **Each node can "reference" content from other nodes, enabling content reusability.** 4. **Flexible information branching and consolidation.** 5. **On-the-fly removal of unnecessary information.** 6. **Effortless comparison of generation results** **across different models in parallel.** 7. ... #### [](https://doc.flowith.io/flow-mode-general-mode/introduction-to-flow-mode#use-case) Use Case: > In Flow mode, you can effortlessly switch between models within the canvas without the need to start a new conversation. For example, you can compare responses from various major models to the question "Which is greater, 9.11 or 9.9?" in the same canvas. Flowith's general mode not only allows for one-click model switching but also presents each model's performance on this question clearly within the canvas, enabling you to compare results more effectively. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FMcQ66DSgfko4BBFfJ55H%252F%25E6%25B5%2581%25E5%25BC%258F%25E4%25BB%258B%25E7%25BB%258D-%25E6%25A1%2588%25E4%25BE%258B-%25E4%25B8%2580%25E9%2594%25AE%25E5%2588%2587%25E6%258D%25A2%25E6%25A8%25A1%25E5%259E%258B.gif%3Falt%3Dmedia%26token%3D2d0f38ff-de1d-45de-b679-c0ec3a5e2c6e&width=768&dpr=4&quality=100&sign=ee453d87&sv=2) One-Click Model Switching > Beyond just switching models, you can freely alternate between modes within a single canvas. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FY9kvFp9ad3phbv8ihMg6%252F%25E6%25B5%2581%25E5%25BC%258F%25E4%25BB%258B%25E7%25BB%258D-%25E6%2596%2587%25E6%259C%25ACmode%25E5%2588%2587%25E6%258D%25A2%25E5%259B%25BE%25E7%2589%2587mode.gif%3Falt%3Dmedia%26token%3Dd7130a37-0dfb-4991-8729-737b26ae2965&width=768&dpr=4&quality=100&sign=470734af&sv=2) Transition from "Text Generation" to "Image Generation" In addition to this, the general mode also supports uploading text files and image files. You can ask questions related to text files and image files. [Advanced Features — Node References](https://doc.flowith.io/flow-mode-general-mode/advanced-features-node-references) * * * We will continuously update additional information on Flow usage and functionality. [PreviousWelcome to flowith](https://doc.flowith.io/) [NextFlow Mode User Guide](https://doc.flowith.io/flow-mode-general-mode/flow-mode-user-guide) Last updated 10 months ago Was this helpful? --- # Welcome to flowith | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2FAHuO48VWVlSzujMHQYk1%2FOriginal5.jpg&width=1248&dpr=4&quality=100&sign=1adf4d93&sv=2) [](https://doc.flowith.io/#what-is-flowith) What is flowith? ----------------------------------------------------------------- Flowith is the next-generation AI creation tool. Unlike traditional chat-based AI tools, Flowith's interaction is based on a two-dimensional canvas that allows you to simultaneously interact with multiple AI models within a single workspace, enabling content generation that significantly boosts efficiency and facilitates entering a flow state during the creative process. Flowith Oracle is our proprietary next-generation AI agent system designed to help you complete multi-step, highly complex tasks and requirements. It supports: * Autonomous planning * Task breakdown and completion * Unlimited tool invocation * Self-optimization and refinement Flowith's **Knowledge Garden** is designed to let you interact with AI using your own knowledge. It automatically organizes, manages, and helps you create with your information. The system breaks down your files, notes, and online resources into small pieces of information called Seeds. Using AI, Flowith connects these Seeds to create a dynamic web of knowledge. When you chat with the AI, it automatically uses the relevant information from your Knowledge Garden, making its responses more accurate and tailored to your needs. Compared to traditional architectures like AutoGPT, **Oracle** demonstrates more mature and stable performance. Additionally, by leveraging flowith's canvas-based interaction, Oracle can better present specific steps and comprehensive information to you. Product Entry: [flowith.io](https://flowith.io/) Official Website: [try.flowith.io](https://try.flowith.io/) Official Twitter: @flowith\_ai Collaboration Contact: [\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection) Discord: [https://discord.com/invite/g95B77xbpD](https://discord.com/invite/g95B77xbpD) [](https://doc.flowith.io/#why-use-flowith) Why use flowith? ----------------------------------------------------------------- **More Efficient Interaction** Compared to the linear dialogue approach of traditional Chat UIs, flowith utilizes a two-dimensional canvas interaction format. This allows you to: * Engage in multi-threaded conversations * Open multiple branches * Better suited for scenarios such as: * Long-form content generation * Comparing AI-generated results * Debugging prompts These features make flowith more adaptable to various complex tasks and creative processes. **Rich Model Selection** Unlike interactions limited to a single type of large language model, our platform offers users a diverse range of models to choose from. This includes: * Multimodal models such as OpenAI: o1 Preview、GPT-4o、4o-mini、Claude 3.5 sonnet * Image generation models like Stable Diffusion and DALL·E 3 This variety enables users to select the most appropriate model for their specific needs and use cases. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FCPnTMkfEEyg8JGWRzOI5%252F%25E6%25AC%25A2%25E8%25BF%258E-%25E4%25B8%25B0%25E5%25AF%258C%25E7%259A%2584%25E6%25A8%25A1%25E5%259E%258B%25E9%2580%2589%25E6%258B%25A9.png%3Falt%3Dmedia%26token%3D11a0265e-e409-4a8b-8422-fda9fce56ff2&width=768&dpr=4&quality=100&sign=741cd1e&sv=2) **Real-time Collaboration and Interactive Features** Flowith's canvas-based interface is inherently suited for multi-user collaboration. Here, you can share a flow link with your colleagues, allowing them to: * Directly view generated results * Join your creative process * Collaboratively generate and debug content ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FpxMq9sMyT6VrhyT60Yzv%252Fmnggiflab-compressed-%25E6%25AC%25A2%25E8%25BF%258E-%25E6%259B%25B4%25E5%25A4%259A%25E7%259A%2584%25E4%25BA%2592%25E5%258A%25A8%25E5%25BD%25A2%25E5%25BC%258F-%25E5%2588%2586%25E4%25BA%25AB.gif%3Falt%3Dmedia%26token%3Da1f24a75-13df-4d2f-b737-3369f72ff638&width=768&dpr=4&quality=100&sign=e45bcb02&sv=2) Copy and Share Links with Your Friends ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FwYg3hPdSM3SsR0RA9Ape%252Fmnggiflab-compressed-%25E6%25AC%25A2%25E8%25BF%258E-%25E6%259B%25B4%25E5%25A4%259A%25E7%259A%2584%25E4%25BA%2592%25E5%258A%25A8%25E5%25BD%25A2%25E5%25BC%258F-%25E8%25AF%2584%25E8%25AE%25BA.gif%3Falt%3Dmedia%26token%3D84f91721-1269-4004-8bc9-8e78e676b267&width=768&dpr=4&quality=100&sign=e023058e&sv=2) Engage in Interactive Discussions Through Comments **Streamlined Inquiry Methods** The Oracle mode eliminates the need for manual prompt engineering. Users can simply input their questions or requirements, and Oracle will autonomously plan, break down, and complete complex tasks. It automatically generates and presents executable, editable, and specific task steps, resulting in more feasible and actionable outcomes. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FRh5exgIxEoGJpUfA7DhI%252Fmnggiflab-compressed-%25E6%25AC%25A2%25E8%25BF%258E-%25E4%25BE%25BF%25E6%258D%25B7%25E7%259A%2584%25E6%258F%2590%25E9%2597%25AE%25E6%2596%25B9%25E5%25BC%258F.gif%3Falt%3Dmedia%26token%3D5ec0fe2f-3356-4a32-acc4-e968c379f51c&width=768&dpr=4&quality=100&sign=8c2e9733&sv=2) Oracle Mode: Automated Task Decomposition **Unlimited Tool Invocation** The Oracle mode supports the integration of external tools (which can be understood as plugins), such as Twitter search, Xiaohongshu search, summarization and reasoning, mind map generation, PowerPoint presentation creation, and webpage generation. It autonomously selects the most appropriate tool from its tool library based on the subtask at hand and invokes it as needed. The number of tools supported by Oracle continues to increase, with its framework allowing for an unlimited number of tool integrations. **Automated Knowledge Organization Powered by AI** Flowith automatically breaks down your uploaded documents, notes, links, and other materials. This process requires no manual categorization or tagging, significantly reducing the time cost of knowledge organization. Additionally, the system automatically establishes connections between knowledge units, creating a structured knowledge network that eliminates information silos and makes knowledge retrieval and utilization more efficient and convenient. **More Precise and Professional Responses** Flowith automatically breaks down materials into AI-comprehensible basic knowledge units called "Seeds." Each Seed contains core concepts or key points. Through AI's deep understanding and retrieval capabilities, these Seeds can be precisely accessed to provide more accurate, contextually relevant support. The system can deeply understand the context of your knowledge base, saying goodbye to "hallucinations" and ineffective responses. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252Fp2WqV5b9dnXSE7hJ6PMq%252FGarden.gif%3Falt%3Dmedia%26token%3D030bf4ee-6e33-421f-accd-fe17dda40b98&width=768&dpr=4&quality=100&sign=d5f454b0&sv=2) **Seamless Creative Experience** Content generated on the canvas can be imported into the Composer with a single click, where you can edit, refine, and polish your work. From capturing inspiration to in-depth research, from initial drafts to final publication, Flowith helps you complete everything in one place. It's not just a storage system but a powerful engine for content creation. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FNtK42svadPnwgmvNXi6v%252Fcomposer.gif%3Falt%3Dmedia%26token%3D58688cbe-246b-486e-bb0e-d4a28700ae29&width=768&dpr=4&quality=100&sign=60063d90&sv=2) * * * We believe that, similar to human-to-human interaction and collaboration, the optimal way for humans and AI to interact should be divergent, multi-threaded, and collaborative. Therefore, we have pioneered an AI interaction method that is completely different from traditional conversational GPT assistants. Collaborate freely with AI on an infinite canvas, allowing for both unlimited ideation and the ability to transform ideas into reusable workflows through clear processes. Now, let's begin with Flo! [](https://doc.flowith.io/#quick-start) **Quick Start** ------------------------------------------------------------ [Flow Mode User Guide](https://doc.flowith.io/flow-mode-general-mode/flow-mode-user-guide) [NextIntroduction to Flow Mode](https://doc.flowith.io/flow-mode-general-mode/introduction-to-flow-mode) Last updated 1 month ago Was this helpful? --- # Flowith Knowledge Base Use Cases | Flowith Official Document Let's explore some impressive use cases collected by our team during beta testing and public release: 1. **Social Media Content Creation — Use Case from Community** Alex, a Xiaohongshu (RED) content creator, built an influencer account through high-quality content. However, his growth plateaued as purely original content couldn't keep up with trending topics. During our beta testing, he uploaded his writing materials and previous articles to the flowith knowledge base. When drafting new content, flowith automatically matched relevant materials and generated personalized text based on his writing style. Instead of starting from scratch each time, Alex now simply inputs a new topic and Flowith rapidly generates high-quality content that aligns with his style and incorporates his personal materials. He reports an "80% time savings" —allowing him to focus solely on brainstorming creative ideas. 1. **Education & Academia (Exam Preparation and Essay Writing) – Public Beta Use Case** Ze was preparing for her final exams. She imported her courseware and reference materials into the Flowith Knowledge Base. The system provided her with relevant knowledge points in real-time while she wrote. Ze found that these features not only increased her essay writing efficiency but also made exam preparation more enjoyable. No more tedious searching through materials—learning became smarter and more efficient. 1. **Team Collaboration – Public Beta Use Case** A marketing team at an educational technology company needs to produce hundreds of marketing pieces periodically for various audiences. Before using Flowith, the team's materials were scattered across different platforms, making retrieval and reuse inefficient. Now, with all marketing materials consolidated within Flowith Knowledge Garden, the AI automatically accesses relevant content and generates output that adheres to the team's style. This has significantly reduced repetitive work and ensured greater consistency in quality and style. 1. **Academic Research — Public Beta Use Case** A university professor in our beta program uses flowith to manage materials collected throughout his academic career. The system automatically aggregated data and literature from diverse sources, supporting his research and providing broader perspectives and deeper insights. 1. **Lesson Planning – Public Beta Use Case** A university instructor faced the challenge of integrating hundreds of papers, lesson plans, and presentations accumulated over five years to prepare for a new semester. After importing these materials into flowith, the AI automatically broke them down into knowledge "Seeds" and established connections between them. Now, by simply entering a topic, flowith intelligently matches relevant content, making lesson preparation significantly more efficient. "This has reduced my preparation time by 60% while actually improving the quality of my lessons," the instructor noted. 1. **Poker — Community Use Case** Xavier, a community user, shared his Poker knowledge base during beta testing. He compiled professional players' game analyses, strategy articles, and real-world examples. The result was a "cheat sheet" that supports learning and provides dynamic, high-quality decision-making assistance during poker games. Its effectiveness demonstrably surpasses that of generic AI solutions. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FEskz605cPJKll8eg4P6a%252F%25E5%25BE%25B7%25E6%2589%2591%25E6%259B%25B4%25E6%2596%25B0.jpg%3Falt%3Dmedia%26token%3De419619a-28ce-432d-890b-ba510e6d6bc3&width=768&dpr=4&quality=100&sign=2934db75&sv=2) 1. **College Applications - Community Use Case** A college consultant compiled a knowledge base of hundreds of successful application essays for prestigious universities like Harvard, MIT, and Yale, and published it on the flowith Marketplace. This knowledge base not only analyzes common characteristics of these successful essays but also tailors essay outlines and generates high-quality initial drafts based on an applicant's background. Within a week of its launch, the knowledge base generated over $800 in revenue. Beyond boosting your personal productivity, Flowith knowledge marketplace offers you the opportunity to share your original knowledge bases with the community and earn revenue. Here, you can publish your crafted, high-quality AI knowledge bases, empowering other users to significantly enhance the quality of AI responses using private knowledge. Currently, dozons of leading content creators and influencers from various fields have accepted our invitation and are building their AI knowledge bases. [PreviousIntroduction to Flowith Knowledge Garden](https://doc.flowith.io/knowledge-garden/introduction-to-flowith-knowledge-garden) [NextHow To Build Your Knowledge Garden](https://doc.flowith.io/knowledge-garden/how-to-build-your-knowledge-garden) Last updated 7 months ago Was this helpful? --- # Fundamentals of Prompt Engineering | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2Ffef0NUW8zbPpdLgMeTJ5%2F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE4.jpg&width=1248&dpr=4&quality=100&sign=f8e39647&sv=2) Generative AI is incredibly useful, but it needs human guidance. In many ways, generative AI can be likened to a highly capable new intern at a company—possessing great potential but requiring clear instructions to perform optimally. The ability to properly guide generative AI is a powerful skill. You can direct generative AI by sending a prompt, which is typically a text-based instruction. A prompt is the input provided to the AI assistant, and it significantly influences the output. An effective prompt should be structured, clear, concise, and purposeful. **How to Craft a Well-Structured Prompt** A structured prompt refers to one that is constructed with a clear logic and organization. For example, if you want the model to generate an article, your prompt might need to include information such as: * The article's main topic * An outline of the article * The desired writing style Let's examine a basic example of a discussion question: > "What are the most pressing environmental issues facing our planet, and what measures can individuals take to help address these problems?" We can transform this into a simple assistant prompt by prefacing it with "Answer the following question:" Copy Answer the following question: What are the most pressing environmental issues facing our planet, and what measures can individuals take to help address these problems? Since the results generated from this prompt are inconsistent, with some responses consisting of only one or two sentences, they are not ideal. A typical discussion answer should comprise multiple paragraphs. An effective prompt should provide **specific instructions for format and content**. To improve consistency and quality, you need to eliminate ambiguity in the language. Here's an improved prompt: Copy Write a highly detailed essay, including introduction, body, and conclusion paragraphs, answering the following questions: What are the most pressing environmental issues facing our planet? What measures can individuals take to help address these problems? The second prompt generated a longer output with an improved structure. The use of the term "essay" in the prompt was intentional, as the assistant understands the definition of an essay and is therefore more likely to produce a coherent, structured response. #### [](https://doc.flowith.io/flow-mode-general-mode/fundamentals-of-prompt-engineering#how-to-improve-quality-and-effectiveness) **How to Improve Quality and Effectiveness** There are several key methods to enhance the quality and effectiveness of prompts: * Clearly define your requirements: The model's output will strive to meet your needs as closely as possible. If your requirements are ambiguous, the output may not meet your expectations. * Use correct grammar and spelling: The model tends to mimic your language style, so if there are issues with your language, the output may reflect those problems. * Provide sufficient context: The model generates output based on the contextual information you provide. If the context is insufficient, it may be challenging to produce the desired results. After crafting effective prompts for discussing issues, you'll need to refine the generated results. This may involve adjusting the output to meet specific constraints, such as word count limitations, or combining concepts from different generated results. A simple method for iteration is to generate multiple outputs and review them to understand the concepts and structures being used. Once you've evaluated the outputs, you can select the most suitable ones and combine them into a coherent response. Another iterative approach is to start small and gradually expand. This requires more than one prompt: an initial prompt to write the first paragraph or two, followed by additional prompts to expand on what has already been written. Here's a potential philosophical discussion question: > "Is mathematics invented or discovered? Explain your answer with careful reasoning." Add this to a simple prompt as follows: Copy Write an opening paragraph answering the following question: Is mathematics invented or discovered? Explain your answer with careful reasoning. I generated several responses and found one I liked: Copy The question of whether mathematics is invented or discovered has been a subject of ongoing debate among philosophers and mathematicians. Some argue that mathematical concepts and principles exist independently of human thought, and thus can be considered discovered rather than invented. For example, the concepts of numbers, ratios, and geometric shapes can be observed in nature and abstracted into systematic study. This is a good starting point. I then used another prompt to expand: Copy I am writing a detailed short essay in response to the following prompt: Is mathematics invented or discovered? Explain your answer using careful reasoning. Here is my current progress: Philosophers and mathematicians have long debated whether mathematics is invented or discovered. Some argue that mathematical concepts and principles exist independently of human thought, and therefore can be considered discovered rather than invented. For example, the concepts of numbers, ratios, and geometric shapes are observed in nature and can be abstracted and studied systematically. Write the next paragraph of my essay. I used this prompt to generate several expansions and selected one I liked, ultimately resulting in the following: Copy The debate between philosophers and mathematicians rages on: Is mathematics invented or discovered? Some argue that mathematical concepts and principles exist independently of human thought, and thus can be considered discovered rather than invented. For example: Concepts of numbers, ratios, and geometric shapes are observed in nature and can be abstracted and studied systematically. On the other hand, there are those who contend that mathematics is largely an invention of the human mind. Their arguments include: Mathematical principles are based on logical reasoning, a characteristic of human thought. For instance, Euclidean geometry is founded on abstract axioms and premises accepted as true without proof. Thus, geometry can be viewed as an invention of human thought rather than a discovery. Similarly, mathematical formulas and equations used to model and predict physical phenomena are also products of human reasoning. By using expanded prompts, we can write incrementally and iterate on each step. This approach is particularly useful in scenarios where **generating higher quality output is required and step-by-step modifications** are desired. [PreviousAdvanced Features — Node References](https://doc.flowith.io/flow-mode-general-mode/advanced-features-node-references) [NextIntroduction to Flowith Knowledge Garden](https://doc.flowith.io/knowledge-garden/introduction-to-flowith-knowledge-garden) Last updated 10 months ago Was this helpful? --- # How To Build Your Knowledge Garden | Flowith Official Document [PreviousFlowith Knowledge Base Use Cases](https://doc.flowith.io/knowledge-garden/flowith-knowledge-base-use-cases) [NextHow To Monetize Your Knowledge Garden](https://doc.flowith.io/knowledge-garden/how-to-monetize-your-knowledge-garden) Last updated 7 months ago Was this helpful? --- # Introduction to Flowith Knowledge Garden | Flowith Official Document [](https://doc.flowith.io/knowledge-garden/introduction-to-flowith-knowledge-garden#why-do-you-need-flowith) **Why do you need Flowith?** ---------------------------------------------------------------------------------------------------------------------------------------------- ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252Fe7Uz0iUHodVIObHRG3Wb%252F1280X1280.PNG%3Falt%3Dmedia%26token%3D8c7f5bca-5849-4dc7-a341-53baaaf760bf&width=768&dpr=4&quality=100&sign=ac771cef&sv=2) As a knowledge worker—whether you're a content creator, student, designer, developer, manager, or any other professionals — you're likely familiar with the power of AI tools. Yet, even experienced users have these common frustrations: 1. **Information Fragmentation:** Your notes, documents, and research are scattered across apps, folders, and browser tabs. This makes it difficult to build a cohesive knowledge base and quickly find the information you need. Instead of seamlessly accessing your knowledge, you're constantly starting from scratch, searching through countless files and piecing together prompts for your AI tools. 2. **Collaboration Silos:** Sharing knowledge within a team is challenging. Locating information buried in months-old documents or deciphering a colleague's individual notes creates communication overhead and inconsistencies in output quality. True collaboration is hindered by these isolated pockets of knowledge. 3. **Generic AI Output:** While large language models (LLMs) are powerful, their ‘general nature’ limits their usefulness for specialized tasks. Lacking context about your specific knowledge and professional domain, they struggle to generate targeted content. This can lead to inaccuracies (‘hallucinations’) and inconsistencies, preventing you from fully relying on their output. People are forced to conform to rigid formats and tools rather than focusing on knowledge itself. This approach feels particularly obsolete in today's Gen AI era. [](https://doc.flowith.io/knowledge-garden/introduction-to-flowith-knowledge-garden#ai-powered-knowledge-management-and-content-creation-introducing-flowith-knowledge-garden) **AI-Powered Knowledge Management and Content Creation: Introducing Flowith Knowledge Garden** ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- To bridge knowledge management and creation in the AI era, Flowith 2.0 introduces **"Knowledge Garden"** - a next-generation knowledge management system powered by AI Agents. It improves the effectiveness and efficiency of AI-assisted content creation while automating personal knowledge management: * **Rapid Personal Knowledge Base Creation:** Instead of starting everything from scratch, simply import your existing materials and files, and Flowith's AI, powered by the Seed system, will automatically deconstruct them into fundamental knowledge units for automated management. You can quickly build your AI-powered knowledge base. * **Automated Matching and Generation:** When you use AI for specific tasks, Flowith automatically matches relevant information from your chosen knowledge base. This transforms your flowith knowledge base into a truly intelligent and dynamic "second brain," facilitating more natural knowledge management and application. * **Connecting Knowledge Management and Content Generation:** Your content is automatically linked, forming an organic and intelligent knowledge network. This ensures accuracy, effectively prevents "AI hallucinations," and significantly enhances output quality. * **Simplified User Experience:** Focus on building your personal knowledge base, not complex prompt engineering. Engage in natural dialogue with the AI, utilize and share community knowledge bases, and dramatically reduce your workload—or even monetize your knowledge. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FgdwwOExr0fSHt13tQHuh%252F%25E7%2583%25AD%25E5%258A%259B%25E5%259B%25BE.PNG%3Falt%3Dmedia%26token%3D9a5712b1-6617-44ff-a8c2-8ddd4e11a866&width=768&dpr=4&quality=100&sign=f83a3cc3&sv=2) The Knowledge Garden combines the technical capabilities of LLMs with the human approach to card-based knowledge management. This allows AI to better understand and utilize private knowledge, helping users generate more targeted, higher-quality content. The Knowledge Garden also serves as a more efficient knowledge management tool for users. In short, if learning and working with your specific knowledge base is like a game, Flowith provides a "cheat code" – an AI-powered knowledge management system that delivers the highest quality AI-generated content available on the market. [PreviousFundamentals of Prompt Engineering](https://doc.flowith.io/flow-mode-general-mode/fundamentals-of-prompt-engineering) [NextFlowith Knowledge Base Use Cases](https://doc.flowith.io/knowledge-garden/flowith-knowledge-base-use-cases) Last updated 7 months ago Was this helpful? --- # Advanced Features — Node References | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) [](https://doc.flowith.io/flow-mode-general-mode/advanced-features-node-references#what-is-a-node) **What is a Node?** --------------------------------------------------------------------------------------------------------------------------- A node is the smallest unit on the Flowith canvas. Each AI response and each of your inputs constitutes a node. [](https://doc.flowith.io/flow-mode-general-mode/advanced-features-node-references#how-to-use-variable-nodes) **How to Use Variable Nodes** ------------------------------------------------------------------------------------------------------------------------------------------------ Variable node references are ideal for frequently changing content (such as text that needs translation). ### [](https://doc.flowith.io/flow-mode-general-mode/advanced-features-node-references#plain-text) **Plain Text** **Inserting a Node** 1. Click the "New input-Node" button in the top-left toolbar and drag it onto the canvas. 2. Modify the node's name. 3. Input the content. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FnUYHcnVCAjgoSCbQsM8Q%252F%25E6%25B5%2581%25E5%25BC%258F-%25E9%25AB%2598%25E7%25BA%25A7%25E7%258E%25A9%25E6%25B3%2595-%25E7%25BA%25AF%25E6%2596%2587%25E6%259C%25AC%25E6%258F%2592%25E5%2585%25A5%25E8%258A%2582%25E7%2582%25B9.gif%3Falt%3Dmedia%26token%3D27953361-2383-4d7e-9f5b-5dc2d543d08f&width=768&dpr=4&quality=100&sign=86989d4f&sv=2) Inserting a Node in the Canvas **Referencing a Node** 1. Move the cursor to the desired insertion point. 2. Click the circular button at the bottom of the canvas. 3. Select "Reference Other Node Content". 4. Click on the node you wish to reference. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FmMEZRD43x0Df3pMhx2HU%252F%25E6%25B5%2581%25E5%25BC%258F-%25E9%25AB%2598%25E7%25BA%25A7%25E7%258E%25A9%25E6%25B3%2595-%25E5%25BC%2595%25E7%2594%25A8%25E8%258A%2582%25E7%2582%25B9.gif%3Falt%3Dmedia%26token%3D4bd197ca-fcde-4fa6-aa2f-0e35b2885393&width=768&dpr=4&quality=100&sign=de5770eb&sv=2) The Canvas Will Dim When You can Select the Node to Reference ### [](https://doc.flowith.io/flow-mode-general-mode/advanced-features-node-references#uploading-pdf-or-images) **Uploading PDF or Images** **Uploading a PDF** 1. Position the cursor at the desired insertion point. 2. Click the circular button located below the canvas. 3. Select "Upload PDF". ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FmWYfPhJIWPJoebtZeL3b%252F%25E6%25B5%2581%25E5%25BC%258F-%25E9%25AB%2598%25E7%25BA%25A7%25E7%258E%25A9%25E6%25B3%2595-%25E4%25B8%258A%25E4%25BC%25A0%25E6%2596%2587%25E6%259C%25AC%25E6%2596%2587%25E4%25BB%25B6.gif%3Falt%3Dmedia%26token%3D534b6fd9-9d38-44f9-b877-13ef3e5587b5&width=768&dpr=4&quality=100&sign=a47b9766&sv=2) **The text file will appear as an "Inserted Node" on the canvas** **Uploading Image Files** 1. Position the cursor at the desired insertion point. 2. Click the circular button located at the bottom of the canvas. 3. Select "Upload Image". ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FYFp4a8Psq3c8EHCfcGBC%252F%25E6%25B5%2581%25E5%25BC%258F-%25E9%25AB%2598%25E7%25BA%25A7%25E7%258E%25A9%25E6%25B3%2595-%25E4%25B8%258A%25E4%25BC%25A0%25E5%259B%25BE%25E7%2589%2587.gif%3Falt%3Dmedia%26token%3D97733b74-ac38-407a-ac6e-450f54706684&width=768&dpr=4&quality=100&sign=3df84c0&sv=2) The Uploaded Image will be Displayed on the Canvas as an "Inserted Node" [PreviousFlow Mode User Guide](https://doc.flowith.io/flow-mode-general-mode/flow-mode-user-guide) [NextFundamentals of Prompt Engineering](https://doc.flowith.io/flow-mode-general-mode/fundamentals-of-prompt-engineering) Last updated 10 months ago Was this helpful? --- # How To Improve Your Prompt | Flowith Official Document [](https://doc.flowith.io/agent-mode-neo/how-to-improve-your-prompt#prompt-enhancement-in-regular-mode) Prompt Enhancement in Regular Mode ----------------------------------------------------------------------------------------------------------------------------------------------- ### [](https://doc.flowith.io/agent-mode-neo/how-to-improve-your-prompt#what-is-prompt-enhancement) What is Prompt Enhancement? **Prompt Enhancement** is a feature in **Regular Mode** that helps you turn your ideas into fully functional, detailed prompts — even if you're not sure how to word them at first. Sometimes your idea is just a spark. Instead of needing to write the perfect AI-ready prompt, you can: 1. Type your idea how you’d say it out loud 2. Enhance it with **Claude 4** 3. Instantly use the improved version with a single click This helps beginners **get results faster** — and teaches everyone how better prompts produce better outputs. ### [](https://doc.flowith.io/agent-mode-neo/how-to-improve-your-prompt#how-to-use-prompt-enhancement) How to Use Prompt Enhancement Change from **Agent Mode** to **Regular Mode** 1. Type a detailed rough idea into the prompt box 2. Press Enter 3. You’ll see a refined, detailed version of your idea 4. Click **Use as Prompt** (bottom-left of the enhanced prompt box) click here for more information: [![Logo](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Ficon%252FGB55uwy2fM8LKvZs3gTa%252FGroup%252039.png%3Falt%3Dmedia%26token%3Dd04bf274-dd4f-4f42-9a78-9cb244afa40c&width=48&height=48&sign=36c4e220&sv=2)Canvas Navigation in Agent Mode | Flowith Official Document](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#swapping-to-regular-mode) ### [](https://doc.flowith.io/agent-mode-neo/how-to-improve-your-prompt#what-happens-when-you-click-use-as-prompt) What Happens When You Click \`Use as Prompt\`? When you click the `**Use as Prompt**` button below an enhanced prompt, the system checks the prompt for **input variables** wrapped in **custom tag format** like: Copy {$SALES_REPORT} this will trigger a popup modal asking you to input your file names e.g. (2024 Annual Sales Report.md) or values, and the system will auto-fill the final prompt before sending it. This small step turns generic prompts into **real, usable instructions** that AI can act on right away — no rewriting needed. ### [](https://doc.flowith.io/agent-mode-neo/how-to-improve-your-prompt#why-we-built-it-this-way) Why We Built It This Way **Prompt Enhancement + Smart Input = Better Products Faster** * **See how AI rewrites your idea** to learn how to write better prompts * **Focus on your idea**, not perfect grammar or structure * **Turn thoughts into working projects** without needing to be a developer Even if you’ve never built anything before, this process gets you from _“I have an idea”_ to _“I’m building something real”_ — and that’s the goal. ### [](https://doc.flowith.io/agent-mode-neo/how-to-improve-your-prompt#final-tip) Final Tip If you ever get stuck, just write **what you want**, even if it's messy, just try and get all your thoughts out and clearly describe how you envision the final product. Use Prompt Enhancement, fill in the blanks if needed, and watch your idea come to life. Review your initial prompt versus the generated one to ensure everything you wanted is still preserved before swapping to Agent Mode to start the task [PreviousAgent Neo Use Cases](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases) [NextSubscriptions and Billing](https://doc.flowith.io/other-questions/subscriptions-and-billing) Last updated 1 month ago Was this helpful? --- # Current Capabilities Supported by Neo | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2FFe0YxPmfFaclcndbPzUc%2FAmazing%2520Mockups.jpeg&width=1248&dpr=4&quality=100&sign=e138a96b&sv=2) The following is a list of external tools currently supported by the Flowith Neo mode. This list is continuously being updated. If you have specific tools you'd like Neo to support, or if you're interested in integrating your product's API with Neo, we welcome you to contact us! **Tool Name** **Tool Function** online\_search Intelligently searches and summarizes web content. read\_single\_webpage Reads the static content of a specific URL. use\_browser Operates a live browser to interact with dynamic or complex websites. search\_images Searches for images online using keywords. search\_twitter Searches for tweets using keywords. watch\_video Retrieves subtitles from a YouTube video. gen\_document Generates text documents like reports and summaries. gen\_image Creates or modifies images from text or image prompts. gen\_3d\_from\_image Generates a 3D model from a single image. gen\_webpage Creates fully functional webpages or apps from scratch. refine\_webpage Modifies or updates an existing webpage project. solve\_logic\_problem Performs rigorous logical reasoning and problem analysis. generate\_storyboard Generates a scene-by-scene storyboard for a video. gen\_video\_segment Generates a single video clip from an image or storyboard scene. gen\_music Generates original music from a text description. gen\_speech Converts text into multilingual speech. ask\_user\_more Asks the user for additional information or confirmation. send\_email Sends task results to the user via email. [PreviousAbout Agent Neo](https://doc.flowith.io/agent-mode-neo/about-agent-neo) [NextAgent Mode User Guide](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide) Last updated 1 month ago Was this helpful? --- # How to Report Issues | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) If you encounter any issues or have suggestions during your use of our product, we welcome your feedback. **Contact us via email:** Official email: [\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection) **Contact us through the product page:** 1. Click on the "Subscribe and Settings" button in the bottom left corner of the product page at flowith.io 2. In the "Subscribe & Settings" menu, select "Feedback". 3. Join Discord [PreviousSubscriptions and Billing](https://doc.flowith.io/other-questions/subscriptions-and-billing) [NextInformation Gathering + Organization Maestro](https://doc.flowith.io/use-case/information-gathering-+-organization-maestro) Last updated 6 months ago Was this helpful? --- # Information Gathering + Organization Maestro | Flowith Official Document You can now have dozens, even hundreds of AI Agents working for you simultaneously. Oracle is a more practical and powerful agent framework than AutoGPT. It autonomously plans tasks, designs workflows, calls tools, and iterates on subsequent workflows. [PreviousHow to Report Issues](https://doc.flowith.io/other-questions/how-to-report-issues) Last updated 10 months ago Was this helpful? --- # Agent Mode User Guide | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2Ffef0NUW8zbPpdLgMeTJ5%2F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE4.jpg&width=1248&dpr=4&quality=100&sign=f8e39647&sv=2) [](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide#getting-started) Getting Started ---------------------------------------------------------------------------------------------------- ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252Foafwm2nT8tjKjyiIyYSP%252FGetting_started.png%3Falt%3Dmedia%26token%3D1bb8d3b0-42f4-4ae8-b2ca-5769e8cffa70&width=768&dpr=4&quality=100&sign=42732ecc&sv=2) Ensure Agent Mode is toggled ON to activate [](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide#agent-mode-interface-guide) Agent Mode Interface Guide -------------------------------------------------------------------------------------------------------------------------- ### [](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide#id-1.-agent-mode-toggle) **1\. Agent Mode Toggle** * **Button:** `AGENT MODE ON/OFF` * **Function:** Enables or disables autonomous task planning, execution, and re-planning. * **Usage:** Click once to toggle the agent state. > When enabled, Neo autonomously handles multi-step tasks using the defined runtime tools and constraints. [](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide#control-panel-buttons) Control Panel Buttons ---------------------------------------------------------------------------------------------------------------- ### [](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide#id-2.-output-length-setting) 2\. Output Length Setting * **Function:** Determine the output Length of Agent Neo's responses * **Usage:** Click the lines on the GUI slider to adjust Output Length ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FZRIb7ZJrk14MRuY55lsQ%252Fcontext_lengthGUI.gif%3Falt%3Dmedia%26token%3D59c0e1f8-500f-4313-a8a1-3fec0e7fb716&width=768&dpr=4&quality=100&sign=18505bf0&sv=2) Select your desired Output Length > You can adjust the maximum output length here to control the upper limit of Credits consumption. Each setting provides example use cases to ensure you can control the consumption of Credits according to your own usage scenarios and achieve the best results. ### [](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide#id-3.-upload-button) 3\. Upload Button * **Function:** Upload external files or documents into the canvas context. * **Usage:** Use this to load task-relevant data into the current canvas session. ### [](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide#id-4.-start-button) 4\. Start Button * **Icon:** Arrow with “start” label * **Function:** Triggers agent execution using current input. * **Usage:** Final step to launch the plan once your input is ready. ### [](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide#id-5.-knowledge-base-toggle) 5\. Knowledge Base Toggle * **Button:** Gray circle next to "Knowledge Base" * **Function:** Turns on/off Knowledge Base integration. * **Usage:** Enable to allow Neo to reference saved documents or context in planning. > Useful when working with previously uploaded documents, structured data, or multi-turn knowledge workflows. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FXqlNHt9M7C1panCk9lyu%252FKnowledgebase_toggle.gif%3Falt%3Dmedia%26token%3D71e8b7bb-4301-4e98-8779-f0d28c3be35a&width=768&dpr=4&quality=100&sign=ae5765ca&sv=2) [](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide#available-tools-for-neo) Available Tools For Neo -------------------------------------------------------------------------------------------------------------------- > You can browse the Available tools you wish to include or exclude in your Prompt for Neo to have available during the session, all credits usage are clearly labeled to help you plan your project ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FGsfQUxKoU8EIuf0fy7Pa%252FAvailable_tools.gif%3Falt%3Dmedia%26token%3D8c134b4b-329c-4557-b653-e54c327358cf&width=768&dpr=4&quality=100&sign=c6197190&sv=2) [](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide#getting-started-1) **Getting Started** ---------------------------------------------------------------------------------------------------------- 1. Enter your request in the input box. 2. Ensure **Agent Mode** is toggled to **ON**. 3. Use the control buttons to configure your agent before running. 4. Click the **Start** button to launch the agent execution. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252Fb4wDCbS1Y5RQ5YNG5Gsk%252FQuick_Start.gif%3Falt%3Dmedia%26token%3D233c55da-d541-49f3-b02f-5a9dd956a56c&width=768&dpr=4&quality=100&sign=a867958b&sv=2) Keep "AGENT MODE" Enabled [](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide#how-to-use-recipe-templates) How to Use Recipe Templates ---------------------------------------------------------------------------------------------------------------------------- 1. Search for desired template keywords in the "Explore" section 2. Click "Use this recipe" to view details 3. Click "Confirm & Run" to apply the template ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FSdg6zWBpFKmmg5uwDR9u%252FOracle-%25E4%25BD%25BF%25E7%2594%25A8%25E6%258C%2587%25E5%258D%2597-%25E5%25A6%2582%25E4%25BD%2595%25E4%25BD%25BF%25E7%2594%25A8%25E6%25A8%25A1%25E7%2589%2588.gif%3Falt%3Dmedia%26token%3Df1afc129-4819-4d4e-858b-4d95e3fcbd8c&width=768&dpr=4&quality=100&sign=e4bf6d92&sv=2) Professional Templates are Available within the Community for Selection [PreviousCurrent Capabilities Supported by Neo](https://doc.flowith.io/agent-mode-neo/current-capabilities-supported-by-neo) [NextCanvas Navigation in Agent Mode](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode) Last updated 1 month ago Was this helpful? --- # About Agent Neo | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2FBOw5ufjnihpyHsRl6OPP%2F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png&width=1248&dpr=4&quality=100&sign=82f401d1&sv=2) [](https://doc.flowith.io/agent-mode-neo/about-agent-neo#what-is-agent-neo) **What is Agent Neo?** ------------------------------------------------------------------------------------------------------- Agent Neo is a cutting-edge intelligent agent designed by the Flowith team to autonomously handle complex, multi-step tasks. It represents an evolution beyond static, pre-planned execution agents, operating with a dynamic and adaptive methodology. Key characteristics of Neo include: * **Autonomous:** Interprets high-level goals and works independently toward a solution. * **Dynamic:** Adapts its approach in real-time based on new information and the results of previous actions. * **Tool-Oriented:** Utilizes a diverse and expanding set of tools to gather information, process data, and execute actions. * **Self-Correcting:** Identifies failures or insufficient results and modifies its strategy to overcome obstacles and achieve the goal. Neo's external capabilities are continuously updated and upgraded, allowing it to become increasingly powerful over time. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2FFe0YxPmfFaclcndbPzUc%2FAmazing%2520Mockups.jpeg&width=768&dpr=4&quality=100&sign=e138a96b&sv=2) Neo Recipe and Nodes In other words, Neo is a **data-level** (operating at the data layer, without physical medium or real-world interaction) **generalized** (capable of performing general tasks, not limited to pre-trained or pre-defined functions) **Agent** (beyond just an LLM, able to utilize tools, manage memory, and collaborate with multiple LLMs) that can complete complex tas**ks** (beyond simple queries or commands) and adapt to new situations (capable of self-learning and processing in novel environments or tasks). [](https://doc.flowith.io/agent-mode-neo/about-agent-neo#core-philosophy-from-static-plans-to-dynamic-recipes) **Core Philosophy: From Static Plans to Dynamic Recipes** ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------- The term "Recipe" originally refers to a set of instructions for preparing a dish. In the context of Neo, a Recipe represents a systematic workflow that AI executes to complete user tasks. When a user inputs a requirement or instruction, Neo breaks down the overall directive into smaller subtasks based on its capabilities and available tools. These subtasks are then passed to a task allocator for step-by-step execution. > A **Recipe** is a living, evolving set of instructions that Neo builds and refines as it works, not just a fixed list created at the beginning. This dynamic approach is more effective for complex problems where the full path to a solution is not knowable in advance, allowing Neo to navigate uncertainty and adapt to new discoveries. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2FBOw5ufjnihpyHsRl6OPP%2F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png&width=768&dpr=4&quality=100&sign=82f401d1&sv=2) A Recipe Generated by Neo [](https://doc.flowith.io/agent-mode-neo/about-agent-neo#the-anatomy-of-a-neo-task-core-concept) **The Anatomy of a Neo Task: Core Concept** ------------------------------------------------------------------------------------------------------------------------------------------------- ### [](https://doc.flowith.io/agent-mode-neo/about-agent-neo#the-dynamic-recipe-neos-evolving-strategy) **The Dynamic Recipe: Neo's Evolving Strategy** A Recipe is the real-time sequence of steps Neo executes. It is generated step-by-step during the task, with each new action informed by the outcome of the previous one. Users can monitor this process through a live execution log, which provides transparency into Neo's "thought process" as it happens. The flow is cyclical and adaptive: Goal -> First Step -> Execution -> Result Analysis -> Next Step -> ... -> Completion ### [](https://doc.flowith.io/agent-mode-neo/about-agent-neo#the-dispatcher-and-tools-neos-capabilities) **The Dispatcher & Tools: Neo's Capabilities** The Dispatcher, also known as a Task Allocator, is a crucial component in the execution of subtasks generated by Neo. For each Neo-generated subtask, a dedicated Agent within the Dispatcher is responsible for allocation and implementation. This Agent identifies the most suitable tool from Neo's supported toolset and then executes and completes the subtask. For example, when the subtask is "**search for user reviews of flowith on social media**", the task allocator will invoke the "**online\_search**" and "**read\_single\_webpage**" to conduct searches for Flowith user reviews. You can view the currently supported tools for Neo on the Neo conversation initiation page. This list is updated in real-time, and we continuously add the most useful tools to Oracle's supported toolkit. [Current Capabilities Supported by Oracle](https://doc.flowith.io/agent-mode-neo/current-capabilities-supported-by-oracle) ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2Ffef0NUW8zbPpdLgMeTJ5%2F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE4.jpg&width=768&dpr=4&quality=100&sign=f8e39647&sv=2) Tools supported by Neo ### [](https://doc.flowith.io/agent-mode-neo/about-agent-neo#intelligent-context-and-memory-management) **Intelligent Context & Memory Management** Complex tasks often generate extensive context. To maintain efficiency and accuracy, Neo employs an intelligent memory manager. For each step, the manager evaluates which information from previous steps is essential and injects only the necessary context. This prevents context overload and ensures the Large Language Model (LLM) can focus on the most relevant data for the current subtask. ### [](https://doc.flowith.io/agent-mode-neo/about-agent-neo#adaptation-and-self-correction) **Adaptation and Self-Correction** Neo's ability to self-correct is central to its power. The entire Recipe is fluid and can be altered mid-execution based on real-time results. This adaptation is triggered automatically by events such as: * **Tool Failure:** If a tool returns an error or no results, Neo will automatically try a different tool or rephrase its query. * **Unsatisfactory Results:** If the information gathered is insufficient, Neo will autonomously add new steps to find more data. * **New Discoveries:** Information found in one step can completely change the subsequent steps in the Recipe. > **Example of Adaptation:** > > 1. **Goal:** "Summarize recent AI hardware announcements." > > 2. **Step 1:** Neo uses > > Google Search > > for "AI hardware announcements 2025". > > 3. **Result:** Results are mixed, including news articles and one press release for a "Project Chimera" chip. > > 4. **Adaptation:** Neo identifies "Project Chimera" as a key term and creates a new sub-task to investigate it. > > 5. **Step 2:** Neo performs a new, targeted search for > > "Project Chimera" chip specifications > [PreviousHow To Monetize Your Knowledge Garden](https://doc.flowith.io/knowledge-garden/how-to-monetize-your-knowledge-garden) [NextCurrent Capabilities Supported by Neo](https://doc.flowith.io/agent-mode-neo/current-capabilities-supported-by-neo) Last updated 1 month ago Was this helpful? --- # How To Monetize Your Knowledge Garden | Flowith Official Document The AI revolution is transforming how we create and monetize knowledge content. As traditional text and video-based content becomes obsolete, AI-powered content creation and distribution is emerging as the new standard. Flowith, the world's most popular canvas-based AI generation platform, is launching its "Knowledge Marketplace" in 2025. Our vision is to become the "OnlyFans" for knowledge creators in the AI era, helping content creators maximize their earnings and exposure while delivering superior value to users. During our beta testing, one of our **partner creators has already seen success**. One creator published an original wealth management knowledge base and attracted over 2,000 users and generated over $3,000 USD equivalent in revenue in just two weeks. ### [](https://doc.flowith.io/knowledge-garden/how-to-monetize-your-knowledge-garden#what-is-flowith-knowledge-marketplace) **What is Flowith Knowledge Marketplace?** ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FC5rpoiKRqNuk9OWaw94m%252FEn%2520MARKET.jpg%3Falt%3Dmedia%26token%3Dc3a2a571-11db-411d-985d-91e40379a787&width=768&dpr=4&quality=100&sign=88fe46f7&sv=2) Large language models (LLMs) are trained on massive datasets and possess extensive knowledge. However, their internal memory is limited, and they can't readily access the most up-to-date, specific, or private information. Flowith solves this problem by connecting LLMs to external knowledge bases. This allows LLMs to dynamically access and retrieve relevant information, minimizing inaccuracies and improving the quality and control of their output. Flowith uses a proprietary AI-powered knowledge management framework to organize and structure information. This framework breaks down uploaded documents and data into individual knowledge items, making them easily searchable and retrievable. When you use flowith with an LLM, these relevant knowledge items are automatically integrated, dramatically improving the accuracy, relevance, and effectiveness of the LLM's responses. Think of Flowith as a knowledge booster for LLMs. By integrating your own private data and your team’s expertise, you can achieve the best possible results from AI generation tools. The Flowith Knowledge Marketplace lets you share your knowledge bases for free or even monetize them. [](https://doc.flowith.io/knowledge-garden/how-to-monetize-your-knowledge-garden#how-to-build-a-good-knowledge-base) How to build a GOOD Knowledge Base? ------------------------------------------------------------------------------------------------------------------------------------------------------------- There are two proven approaches to creating a knowledge base that provides value to users and generates revenue: 1. **Original Creation** If you're already a content creator, blogger, or subject matter expert, Flowith offers a new avenue to monetize your expertise with AI. Simply upload your original content in various formats to share and publish it on the marketplace. 1. **Content Curation** In today's information-rich world, access to quality content isn’t the problem; finding and organizing the most valuable insights is. With flowith, you can curate and compile online resources around a specific topic, build a knowledge base, and begin monetizing your curated collection. [](https://doc.flowith.io/knowledge-garden/how-to-monetize-your-knowledge-garden#source-upload-guidelines) Source Upload Guidelines ---------------------------------------------------------------------------------------------------------------------------------------- ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FCIG2HQ22vnzTqA0lFUwU%252F2.jpg%3Falt%3Dmedia%26token%3D6bbdce7a-dc55-4f18-9d29-04167389ee3a&width=768&dpr=4&quality=100&sign=9749476f&sv=2) * **Length Requirements:** * Content length is flexible - responses can be generated for any text containing relevant keywords * Text summarization is not supported * Source texts exceeding 1,000 characters will be automatically segmented * **File Format Specifications:** * No preferred format restrictions, but PDF image recognition is not supported * Images must be uploaded separately; images embedded in long texts will be ignored * Code reading/processing is not supported * **Supported Content Types:** * Standalone image uploads are functional * Text-based content is fully supported * Long-form content requires segmentation [](https://doc.flowith.io/knowledge-garden/how-to-monetize-your-knowledge-garden#how-to-publish-and-market-your-knowledge-base-in-the-marketplace) How to Publish and Market Your Knowledge Base in the Marketplace? ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FaysanpBX6RC9lR7t2Sxv%252F%25E5%25B8%2582%25E5%259C%25BAEn2.jpg%3Falt%3Dmedia%26token%3D800d5639-8577-41f7-b327-9fd0d76af7a5&width=768&dpr=4&quality=100&sign=a80e9542&sv=2) Creating a great knowledge base is just the first step. Knowledge Marketplace is the platform where users browse and purchase knowledge bases. With Flowith's large community actively searching for solutions to their problems, a clear and compelling title and description can help your knowledge base stand out and attract more users—ultimately increasing your potential earnings. [](https://doc.flowith.io/knowledge-garden/how-to-monetize-your-knowledge-garden#how-to-write-your-knowledge-base-description) How to Write Your Knowledge Base Description -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ### [](https://doc.flowith.io/knowledge-garden/how-to-monetize-your-knowledge-garden#creating-a-great-title) Creating a Great Title * **Single Source Knowledge Bases:** If your knowledge base draws from a single source, we recommend using the source's name directly (e.g., "The Art of War"). * **Collections, Series, and Composite Knowledge Bases:**If your knowledge base compiles information from multiple sources, choose a title that clearly and concisely describes the overall theme or topic. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FkovzIf6O8LbzxBIoXCRZ%252F1.jpg%3Falt%3Dmedia%26token%3Dbd246f7c-7c81-4937-b60c-8a12ed99fabf&width=768&dpr=4&quality=100&sign=e45bac13&sv=2) ### [](https://doc.flowith.io/knowledge-garden/how-to-monetize-your-knowledge-garden#well-written-description) Well-written description ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252F5OnG3s9BlzxEOTtplzat%252F3.jpg%3Falt%3Dmedia%26token%3D87beaade-16f1-4b6b-85a6-a24addbb14d0&width=768&dpr=4&quality=100&sign=c15bd3fe&sv=2) Your knowledge base description is your storefront - it helps users quickly understand its value and use case. Based on experience, we recommend addressing these four key aspects: 1. **Source Information** * Clearly state content sources * Indicate information timeliness and update frequency * Include author or maintenance team credentials 1. **Key Features** * Outline the main content scope * Explain how information is organized * Highlight unique advantages or characteristics 1. **Target Audience** * Define intended user groups * Describe typical use cases 1. **Value Proposition** * Specify which problems it solves * Explain user benefits * Describe expected learning or practical outcomes Copy Example： This REACT best practices knowledge base is designed for front-end developers. Based on our team's three years of practical project experience and updated monthly, it provides systematic component development guidelines and performance optimization tips. This knowledge base helps developers improve code quality and boost development speed. Key Takeaways: Keep your description concise and clear, enabling users to quickly determine if it's right for them. Ensure accuracy and avoid overpromising. [PreviousHow To Build Your Knowledge Garden](https://doc.flowith.io/knowledge-garden/how-to-build-your-knowledge-garden) [NextAbout Agent Neo](https://doc.flowith.io/agent-mode-neo/about-agent-neo) Last updated 7 months ago Was this helpful? --- # Agent Neo Use Cases | Flowith Official Document [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#what-types-of-problems-is-neo-agent-mode-suitable-for-solving) What Types of Problems Is Neo (Agent Mode) Suitable for Solving? ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Agent Neo is not a static chat assistant—it’s a dynamic agent designed to **execute**, **adapt**, and **deliver outcomes** across complex, multi-step workflows. When used in Agent Mode, Neo becomes a structured operator that: * Understands your intent — even from vague prompts * Plans each step using tool-driven logic * Executes real actions (file generation, searches, code output) * Responds to results in real time by refining its plan * Finishes with professional-grade deliverables or previews This makes Agent Mode ideal for advanced automation, technical project work, and any goal where **step-by-step reasoning** must be paired with **tool usage, file output, and task persistence.** [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#neo-agent-mode-is-ideal-for-tasks-that) Neo (Agent Mode) is ideal for tasks that: --------------------------------------------------------------------------------------------------------------------------------------------------- **Refine Broad Questions** You present a broad question, and Neo follows up with targeted inquiries to narrow and specify your goals. * **Plan & Execute Complex Workflows** Independently break down tasks into actionable steps, generate a concrete execution plan, and carry it out—from start to finish. * **Automate Multi-Step Processes** Build batch scripts, pipelines, or orchestrators (e.g. “Create a Tailwind-powered webpage, remove unused files, run tests, and package as a ZIP.”) * **Research & Report Generation** Conduct in-depth searches across platforms, compile findings into structured reports, and even produce professional web pages or PowerPoint decks to present your results. * **Leverage Community Templates** Search, view, and apply community-provided templates (webpage layouts, slide themes, code snippets) to accelerate development. * **Integrate Tools & Data** Call APIs, scrape and summarize web content, query databases, and perform file I/O directly within your prompt. * **Maintain State Across Steps** Remember variables, previous outputs, and user preferences to build guided tutorials, chatbots, or form-fillers that “know” where you left off. * **Compose Structured, Long-Form Outputs** Create whitepapers, technical guides, multi-section proposals, or any document requiring consistent style and formatting. * **Solve Niche or Domain-Specific Tasks** From custom asset packers to conditional logic flows (“If X, then run Y, else run Z”), Neo adapts to specialized requirements. * **Orchestrate Other AI Models** Chain multiple AI services—e.g., one for search embeddings, another for summarization, a third for code generation—seamlessly in one workflow. [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#what-can-neo-agent-mode-help-build-or-automate) What can Neo (Agent Mode) Help Build or Automate? ------------------------------------------------------------------------------------------------------------------------------------------------------------------- With the ability to plan workflows, run tools, and output complete files or web experiences, Neo can handle everything from publishing pipelines to full site builds. Here are examples of what Neo can generate or orchestrate: ### [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#content-and-documentation) Content & Documentation * Create structured **knowledge wikis**, FAQ pages, or help centers * Write and format **technical documentation** with consistent style * Summarize multi-source input into **long-form reports or whitepapers** ### [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#web-and-frontend-development) Web & Frontend Development Build **static websites** with Tailwind, HTML, and JS (ZIP ready) * Scaffold **landing pages**, personal portfolios, or product showcase sites * Inject custom animations, form logic, and scroll behaviors * Create a **static site image generator** or visual content tool ### [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#automation-and-utilities) Automation & Utilities Send **automated emails** using SMTP credentials or mocked endpoints * Schedule daily or conditional tasks using **script logic or CRON syntax** * Develop **pipelines** (e.g. generate → test → package → export flow) * Parse spreadsheets, convert formats, or generate structured data for reuse ### [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#creative-and-visual-projects) Creative & Visual Projects * Generate original **3D models from text descriptions** or reference images * Output **.glb** formats for use in design pipelines * Launch a **3D scene editor** or sandbox viewer directly in a webpage * Create downloadable assets like icons, vector packs, or themed UI kits > This capability comes from Neo’s ability to interpret your prompt, run a structured plan, use context-aware toolchains, and deliver files or previews—all within a single Flow session. [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#when-to-use-agent-mode-vs.-regular-mode) When to Use Agent Mode vs. Regular Mode -------------------------------------------------------------------------------------------------------------------------------------------------- **Choose Neo (Agent Mode)** **Choose Regular Mode** You need **tool calls**, file I/O, or web searches embedded in your AI workflow. You’re **brainstorming a project roadmap** with another model (e.g., Claude 4) before switching to Neo. You want the AI to **refine** your question through follow-up prompts when details are missing. You’re **drafting a detailed prompt** to describe a promotional video for your webpage’s content. You require the AI to **plan, break down, and execute** multi-step tasks with a concrete plan. You’re **debugging your webpage**—get clear error explanations so you can feed fixes back into Neo. You’re automating **developer tasks** (webpage scaffolding, build scripts, test runners). You want to draft **initial specs** or deep-dive into architecture before automating with Neo. You need **formatted, multi-section outputs** (reports, docs, slides, decks). You’re planning a one-off example (e.g., generating a storyboard prompt, evaluating design directions). You need **conditional logic** or stateful flows (“If X, then Y, else Z”). You’re experimenting with **tool-specific workflows** (e.g., exploring veo3 video prompts manually). Learn about swapping to Regular Mode here: [![Logo](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Ficon%252FGB55uwy2fM8LKvZs3gTa%252FGroup%252039.png%3Falt%3Dmedia%26token%3Dd04bf274-dd4f-4f42-9a78-9cb244afa40c&width=48&height=48&sign=36c4e220&sv=2)Canvas Navigation in Agent Mode | Flowith Official Document](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#swapping-to-regular-mode) [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#frame-your-prompt) Frame Your Prompt ------------------------------------------------------------------------------------------------------ ### [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#deliverable-webpage) Deliverable Webpage Copy Create a modern, mobile-friendly webpage that includes: 1. A large headline with a short subheading at the top. 2. Three sections to show off different features or benefits. 3. A rotating testimonial section with customer quotes. 4. A contact form with name, email, and message fields. 5. A top menu that stays visible while scrolling. The design should feel modern, easy to read, with soft colors, smooth layout, and clear sections that feel inviting and professional. ### [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#sales-report-with-visuals) Sales Report with Visuals Copy Generate a professional sales report from the attached file that includes: 1. A title and short summary at the top. 2. A clean line graph showing sales by month for the past year. 3. A table breaking down totals by product category. 4. A short paragraph analyzing trends or seasonal patterns. 5. A bullet list of recommended next steps based on the data. The report should look clean and organized, with clear labels, legible fonts, and soft colors that make the visuals easy to read in a presentation. ### [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#id-3d-model-of-a-product) 3D Model of a Product Copy Create a realistic 3D model of a reusable water bottle based on this description: - Made of stainless steel with a matte finish - Has a flip-open lid with a locking mechanism - Includes a measurement scale on the side - Slightly curved ergonomic shape The model should be clean and well-proportioned, suitable for use in a product showcase or digital mockup. > You can also Swap to Prompt Enhance mode via the Agent Mode toggle button on the input prompt field located at the bottom of the canvas and enhance your prompts with Claude 4 Opus. Explore our catalog of Flows developed from users like you here: [![Logo](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fflowith.io%2Fapple-touch-icon.png&width=20&dpr=4&quality=100&sign=93d250ae&sv=2)flowith 2.0 - Your AI Creation Workspace, with Knowledgeflowith](https://flowith.io/community?type=oracle) [](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases#quick-start) Quick Start ------------------------------------------------------------------------------------------ **Pick Your Mode** Copy - **Neo (Agent Mode)** when you need automated, stateful, tool-integrated workflows. - **Regular Mode** when you’re prototyping, brainstorming, or need specialized, one-off guidance. 2\. **Use the Right Keyword** Copy - Include **“webpage”** for Neo-driven deliverables. - Include Keywords like **“visualize sales trends”**, **“build performance dashboard”**, or **“generate executive report”** to ensure Neo delivers structured, actionable results tailored to your objective 3\. **Run & Refine** Copy - Start with the concise examples. - Include file names, tool names, or formatting instructions to tailor Neo’s output to your exact needs. [PreviousCanvas Navigation in Agent Mode](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode) [NextHow To Improve Your Prompt](https://doc.flowith.io/agent-mode-neo/how-to-improve-your-prompt) Last updated 1 month ago Was this helpful? --- # Prompt Engineering 基础 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FIoGBv6KmDuBGVIrYtsfM%252F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE4.jpg%3Falt%3Dmedia%26token%3Dc51d1b64-dc29-4d04-a197-e95cb7aa7124&width=1248&dpr=4&quality=100&sign=bbf8cc9c&sv=2) 生成式 AI 非常有用，但它需要人类指导。通常情况下，生成式 AI 能就像公司新来的实习生一样，非常有能力，但需要清晰的指示才能做得好。能够正确地指导生成式 AI 是一项非常强大的技能。你可以通过发送一个 prompt 来指导生成式 AI，这通常是一个文本指令。Prompt 是向助手提供的输入，它会影响输出结果。一个好的 Prompt 应该是结构化的，清晰的，简洁的，并且具有指向性。​ **如何写好一个结构化 prompt** `结构化 prompt 是指 prompt 的构造应该有明确的逻辑和结构。例如，如果你想让模型生成一篇文章，你的 prompt 可能需要包括文章的主题，文章的大纲，文章的风格等信息。` ​让我们看一个基本的讨论问题的例子: > _"我们星球面临的最紧迫的环境问题是什么，个人可以采取哪些措施来帮助解决这些问题？"_ 我们可以将其转化为简单的助手提示，将回答以下问题：放在前面。 Copy 回答以下问题: 我们星球面临的最紧迫的环境问题是什么，个人可以采取哪些措施来帮助解决这些问题？ 由于这个提示生成的结果并不一致，有些只有一两个句子。一个典型的讨论回答应该有多个段落，因此这些结果并不理想。一个好的提示应该给出**具体的格式和内容指令**。您需要消除语言中的歧义以提高一致性和质量。这是一个更好的提示。 Copy 写一篇高度详细的论文，包括引言、正文和结论段，回答以下问题: 我们星球面临的最紧迫的环境问题是什么， 个人可以采取哪些措施来帮助解决这些问题？ 第二个提示生成了更长的输出和更好的结构。提示中使用 “论文” 一词是有意的，因为助手可以理解论文的定义，因此更有可能生成连贯的、结构化的回答。 #### [](https://doc.flowith.io/cn/flowith-normal/prompt-engineering-ji-chu#ru-he-ti-sheng-qi-zhi-liang-he-xiao-guo) 如何提升其质量和效果 提升 prompt 质量和效果的方法主要有以下几点： * 尽量明确你的需求。模型的输出会尽可能满足你的需求，所以如果你的需求不明确，输出可能会不如预期。 * 使用正确的语法和拼写。模型会尽可能模仿你的语言风格，所以如果你的语言风格有问题，输出可能也会有问题。 * 提供足够的上下文信息。模型会根据你提供的上下文信息生成输出，所以如果你提供的上下文信息不足，可能无法生成你想要的结果。 在为讨论问题制定有效的提示后，您现在需要细化生成的结果。这可能涉及到调整输出以符合诸如字数等限制，或将不同生成的结果的概念组合在一起。 迭代的一个简单方法是生成多个输出并查看它们，以了解正在使用的概念和结构。一旦评估了输出，您就可以选择最合适的输出并将它们组合成一个连贯的回答。另一种迭代的方法是逐步开始，然后**逐步扩展**。这需要不止一个提示：一个起始提示，用于撰写最初的一两段，然后是其他提示，以扩展已经写过的内容。以下是一个潜在的哲学讨论问题: > _"数学是发明还是发现？用仔细的推理来解释你的答案。"_ 将其添加到一个简单的提示中，如下所示: Copy 写一个起始段回答以下问题: 数学是发明还是发现？用仔细的推理来解释你的答案。 我生成了一些回答，找到了一个我喜欢的： Copy 哲学家和数学家之间争论不休，数学是发明还是发现。 一些人认为，数学的概念和原则独立于人类思维存在，因此可以被认为是被发现而不是被发明的。 例如，数字、比率和几何形状的概念在自然界中被观察到，可以被抽象化和系统化地研究。 是一个不错的开端。然后我使用另一个提示来扩展： Copy 我正在撰写一篇详细的短文，回答以下提示: 数学是发明还是发现？用仔细的推理来解释你的答案。这是我目前的情况: 哲学家和数学家之间争论不休，数学是发明还是发现。 一些人认为，数学的概念和原则独立于人类思维存在，因此可以被认为是被发现而不是被发明的。 例如，数字、比率和几何形状的概念在自然界中被观察到，可以被抽象化和系统化地研究。写下我文章的下一段。 我使用此提示生成了几个扩展，并选择了一个我喜欢的，最终得到了以下结果： Copy 哲学家和数学家之间争论不休，数学是发明还是发现。 一些人认为，数学的概念和原则独立于人类思维存在，因此可以被认为是被发现而不是被发明的。 例如，数字、比率和几何形状的概念在自然界中被观察到，可以被抽象化和系统化地研究。 另一方面，有人认为数学在很大程度上是人类思维的发明。 数学原理基于逻辑推理，这是人类思维的一个特征。 例如，欧几里德几何基于抽象的公理和前提，被接受为真实而无需证明。 因此，可以认为几何是人类思维的发明，而不是一种发现。 同样，数学公式和方程用于模拟和预测物理现象，这又是人类推理的结果。 使用扩展提示，我们可以逐步地写作并在每个步骤上进行迭代。这对于需要**生成更高质量的输出并希望逐步修改**的情况非常有用。 [Previous高级玩法 — 节点引用](https://doc.flowith.io/cn/flowith-normal/gao-ji-wan-fa-jie-dian-yin-yong) [Next知识花园功能简介](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-gong-neng-jian-jie) Last updated 11 months ago Was this helpful? --- # Subscriptions and Billing | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) [](https://doc.flowith.io/other-questions/subscriptions-and-billing#subscription-plans) **Subscription Plans** ------------------------------------------------------------------------------------------------------------------- We offer highly competitive subscription plans that give you access to cutting-edge AI models from industry leaders including OpenAI, Anthropic, Google, DeepSeek, and Meta. Our service also includes premium models for image and video generation capabilities. For details, please visit: [https://flowith.io/sub](https://flowith.io/sub) [](https://doc.flowith.io/other-questions/subscriptions-and-billing#super-model-oracle) **Super Model — Oracle** --------------------------------------------------------------------------------------------------------------------- In Oracle mode, **each step consumes one T-1 quota**. For example, a task requiring six steps will use 6 T-1 credits. [](https://doc.flowith.io/other-questions/subscriptions-and-billing#important-information) **Important Information:** -------------------------------------------------------------------------------------------------------------------------- ### [](https://doc.flowith.io/other-questions/subscriptions-and-billing#knowledge-base-token-quota-and-usage-management) **Knowledge Base Token Quota and Usage Management** 1. Files uploaded to the knowledge base only count toward your storage quota and won't incur any additional model usage charges. 2. When you engage in conversations using the knowledge base, you will not consume any tokens. Rates depend on your chosen service tier (T1 or T2) and align with our standard conversation service pricing. 3. For detailed information about quota for different subscription plans, please visit: [https://flowith.io/sub](https://flowith.io/sub) **\*Clarification** : We have noticed that some users are creating multiple Flowith accounts to circumvent limits and take advantage of our free tier to access our premium AI models. This abuse of our system negatively impacts the experience of other users. If you have legitimate needs that exceed the standard usage limits, please reach out to our support team [\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection#74070104041b06003412181b031d001c5a171b) . ### [](https://doc.flowith.io/other-questions/subscriptions-and-billing#refund-policy) Refund Policy * Subscription fees are non-refundable * Refunds are available only if you meet the following conditions: * Contact us within 48 hours after payment. * You've made fewer than 200 credits throughout your account lifetime after subscription. **Contact [\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection) :** * For any outstanding issues or special circumstances * For refund requests that meet our criteria * For technical difficulties with cancellation **Contact Information:** Email: [\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection#8cfff9fcfce3fef8cceae0e3fbe5f8e4a2efe3) Discord: [https://discord.com/invite/g95B77xbpD](https://discord.com/invite/g95B77xbpD) [PreviousHow To Improve Your Prompt](https://doc.flowith.io/agent-mode-neo/how-to-improve-your-prompt) [NextHow to Report Issues](https://doc.flowith.io/other-questions/how-to-report-issues) Last updated 2 months ago Was this helpful? --- # 欢迎使用 flowith | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252F66tYVPeBXnhJqTesILCy%252FOriginal5.jpg%3Falt%3Dmedia%26token%3Dbd57e8c9-a939-433d-a94c-73533e385891&width=1248&dpr=4&quality=100&sign=cd0747a0&sv=2) flowith - AI for Deep Work [](https://doc.flowith.io/cn#shen-me-shi-flowith) 什么是 flowith ------------------------------------------------------------------ flowith 是下一代 AI 生产力工具。不同于传统聊天式 AI 工具，flowith 基于二维画布的交互让可以你在一个画布中与多个 AI 模型同时交互，并进行内容生成，让你在创造内容时大幅度提升效率、更容易地进入心流状态。 flowith Oracle 是我们自主打造的下一代 AI agent 系统，它可以帮助你完成多步骤、超复杂的任务和需求。支持自主规划、拆分和完成任务、无限的工具调用、和自我优化完善，比传统的 AutoGPT 等架构有更成熟和稳定的表现。同时，利用 flowith 画布式的交互，Oracle 可以更好的为你呈现具体步骤和全面信息。 flowith 「Knowledge Garden」 知识花园，围绕“让 AI 与您的知识自然互动”这一核心理念进行打造。它结合了自动化知识整理、结构化管理和创作协同，能让你的信息资产真正活跃起来，成为可持续生长的“活知识”。它能够自动将你的文件、笔记和在线资源转化为 AI 可理解的最小知识单元（Seeds 知识种子）。系统通过 AI 智能分析，自动建立知识间的关联，形成一个动态生长的知识网络。在与 AI 对话时，相关的知识会被自动调用，大幅提高 AI 回复的准确性和针对性。不仅如此，通过 Knowledge Marketplace 功能可以在社区中分享和交易高质量的知识库，让知识价值得到更广泛的流通。 产品入口：[flowith.io](https://flowith.io/) 官网：[try.flowith.io](https://try.flowith.io/) 官方推特：@flowith\_ai [https://twitter.com/flowith\_ai](https://twitter.com/flowith_ai) 合作联系邮箱: [\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection) Discord: [https://discord.com/invite/g95B77xbpD](https://discord.com/invite/g95B77xbpD) 注：受合规性限制，flowith 产品不支持在中国大陆网络下直接访问，需要使用 VPN 或切换网络进行正常使用。 [](https://doc.flowith.io/cn#wei-shen-me-yao-shi-yong-flowith) 为什么要使用 flowith ---------------------------------------------------------------------------------- **更高效的交互方式** 相比于传统的 Chat UI 线性的对话方式，flowith 使用二维画布的交互形式，让你可以进行多线程的对话、开启多个分支，更适用于长内容生成、对比 AI 生成结果、调试 Prompt 等场景。 **丰富的模型选择** 相比于单一模型的使用，flowith 为用户提供了种类丰富的模型，比如 OpenAI: o1 Preview、GPT-4o、4o-mini、Claude 3.5 sonnet 等 AI 模型，Stable Diffusion、DALL·E 3 等图片生成模型。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252F9MnfXkbJvbHiG1kOEk7k%252F%25E6%259B%25B4%25E5%25A4%259A%25E6%25A8%25A1%25E5%259E%258B.png%3Falt%3Dmedia%26token%3D276fa58c-6c60-436c-bdd0-9fff15fc1d1a&width=768&dpr=4&quality=100&sign=e59e2155&sv=2) **多人实时协作等互动形式** 基于画布的交互让 flowith 天然适合多人协作。在这里，你可以分享一个 flow 的链接给好友，让他们直接浏览生成结果，也可以加入你的创作，让你们共同进行内容生成与调试。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252Fwfi2CmuqpKqs97H4Ipyl%252Fmnggiflab-compressed-%25E5%2588%2586%25E4%25BA%25AB1.gif%3Falt%3Dmedia%26token%3Df1694f8e-2baa-459f-9363-3f7dbe5d3ac0&width=768&dpr=4&quality=100&sign=aadc7bc9&sv=2) 复制链接分享给朋友 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252F8eUUA5NKzv8vcdsRAXPT%252Fmnggiflab-compressed-%25E8%25AF%2584%25E8%25AE%25BA.gif%3Falt%3Dmedia%26token%3D2509b16b-f0c6-4677-9125-2da39defba92&width=768&dpr=4&quality=100&sign=f97bcd&sv=2) 和朋友进行评论互动 **便捷的提问方式** Oracle 模式不需要输入 Prompt ，只需输入你的问题或需求，Oracle 可以自主规划、拆分并完成复杂任务，自动生成并呈现可执行的、可编辑的任务步骤，最终为你提供一个更完整高质量的结果。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FxKdAMjkQvuiMZ7MfY1NZ%252F%25E6%258B%2586%25E8%25A7%25A3%25E6%25AD%25A5%25E9%25AA%25A4.gif%3Falt%3Dmedia%26token%3D49984a06-636e-400f-a2aa-daa5c1cbe8b6&width=768&dpr=4&quality=100&sign=9563b131&sv=2) Oracle 模式自动拆分任务 **引导式的追问** 相比于传统的 AI 工具不会让用户补充必要信息、生成结果过于笼统模糊等缺点，Oracle 模式可以通过主动追问的方式，一步步获得完整的必要信息，并进行提供一个高质量且深度的最终结果。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FkO307Tdwqo3cJduHpyEY%252F%25E5%25BC%2595%25E5%25AF%25BC%25E5%25BC%258F%25E8%25BF%25BD%25E9%2597%25AE.gif%3Falt%3Dmedia%26token%3D895a6eca-aa1f-47d7-8422-54743d4de1c9&width=768&dpr=4&quality=100&sign=ef4e75be&sv=2) 引导式的追问丰富更多信息 **无限的调用工具** Oracle 模式支持调用外部工具（可理解成插件），比如推特搜索、小红书搜索、总结和推理、生成思维导图、生成 PPT、生成网页等。它支持根据子任务自主从工具库里的选择最合适的工具并进行调用。Oracle 所支持的工具数量也在持续提升，它的框架允许有无限数量的工具接入。 * * * 我们认为，像人与人的互动和协作一样，人和 AI 最佳的互动方式应该是发散、多线程、和相互协作的，因此，我们首创性地设计了和传统对话式 GPT 助手完全不同的 AI 互动方式，通过一个无限的画布，让你和 AI 可以更自由的协作，既可以无限发散，也可以通过清楚的流程把它转化成可复用的工作流。 现在，开始 Flo！ [](https://doc.flowith.io/cn#kuai-su-kai-shi) 快速开始 ------------------------------------------------------- [流式聊天使用指南](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-shi-yong-zhi-nan) [Next流式聊天模式介绍](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-mo-shi-jie-shao) Last updated 2 months ago Was this helpful? --- # 流式聊天模式介绍 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FpE4KsV6mTBBcTyNenIcT%252F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE2.png%3Falt%3Dmedia%26token%3D46e9cddf-7c56-46a7-8c28-57dd7d455231&width=1248&dpr=4&quality=100&sign=ccda9d65&sv=2) 「流式聊天模式」 (flow 模式) 是我们为你打造的下一代和 AI 交互的方式，它是 flowith 的常规交互方式。flowith 通过一个无限大的画布，让你可以同时使用多个 AI，并生成你所需要的内容。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FpE4KsV6mTBBcTyNenIcT%252F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE2.png%3Falt%3Dmedia%26token%3D46e9cddf-7c56-46a7-8c28-57dd7d455231&width=768&dpr=4&quality=100&sign=ccda9d65&sv=2) **为什么要使用 Flow 模式？** 如果你只是需要让 AI 为你做简单的翻译，或者基础的问答，也许普通的聊天模式即可满足你。但如果你需要和 AI 进行多轮问答，并修改之前的输入，或者需要多个 AI 一同为你生成，Flow 模式即可提供完全不同的创作体验。 **Flow 模式具体的优势：** 1. **可以多 AI 同时、多线程进行生成。** 2. **你可以反复修改 AI Prompt，并多次生成。** 3. **每个节点可以 「 引用 」 其他节点中的内容，让内容可复用** 4. **可以对信息随时进行分叉，或汇总** 5. **随时删减不必要的信息** 6. **可以轻松地对比不同模型的生成效果，让它们同时并行生成内容** 7. ... #### [](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-mo-shi-jie-shao#an-li) 案例: > 在这个模式下，你可以在画布中轻松切换模型，完全不需要重开一个新对话。比如在同一画布中对比各大模型对 「9.11 和 9.9 哪个大？」 这个问题的回复。flowith 的常规模式既可以一键切换模型，又可以将各模型的对这个问题的表现在画布中一目了然的呈现，让你更好地对结果进行比对。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FIQl61EwHXGy57Iag0CQI%252F%25E5%25B8%25B8%25E8%25A7%2584%25E6%25A8%25A1%25E5%25BC%258F%25E5%258F%25AF%25E4%25BB%25A5%25E9%2597%25AE%25E4%25BB%2580%25E4%25B9%25881.1.gif%3Falt%3Dmedia%26token%3Da8e45975-4126-4e36-9d14-2ac88d7bda1f&width=768&dpr=4&quality=100&sign=91d5bc4a&sv=2) 一键切换模型 > 不仅仅是切换模型，在一个画布里模式也可以随意切换。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252Fp1vHriL6NFlSg5Z4aKUp%252F%25E4%25BB%258E%25E7%2594%259F%25E6%2588%2590%25E6%2596%2587%25E6%259C%25AC%25E5%2588%25B0%25E7%2594%259F%25E6%2588%2590%25E5%259B%25BE%25E7%2589%2587%25E7%259A%2584%25E8%25BD%25AC%25E6%258D%25A2.gif%3Falt%3Dmedia%26token%3D9ba2f407-f67c-4a77-9932-c57a5eda2dc2&width=768&dpr=4&quality=100&sign=ddbcc9e4&sv=2) 「生成文本」 到 「生成图片」 的转换 除此之外，常规模式也支持文件和图片文件的上传，你可以提问和文件、图片相关的问题。 [高级玩法 — 节点引用](https://doc.flowith.io/cn/flowith-normal/gao-ji-wan-fa-jie-dian-yin-yong) * * * 更多关于 Flow 的用法我们将持续更新。 [Previous欢迎使用 flowith](https://doc.flowith.io/cn) [Next流式聊天使用指南](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-shi-yong-zhi-nan) Last updated 11 months ago Was this helpful? --- # 高级玩法 — 节点引用 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) [](https://doc.flowith.io/cn/flowith-normal/gao-ji-wan-fa-jie-dian-yin-yong#shen-me-shi-jie-dian) 什么是节点 ------------------------------------------------------------------------------------------------------------ 节点是 Flowith 画布中的最小单位。每一个 AI 回复，每一个你的输入，都是一个节点。 [](https://doc.flowith.io/cn/flowith-normal/gao-ji-wan-fa-jie-dian-yin-yong#ru-he-shi-yong-bian-liang-jie-dian) 如何使用变量节点 ----------------------------------------------------------------------------------------------------------------------------- 变量节点引用适用于频繁改变的内容（比如翻译中需要翻译的文本）。 ### [](https://doc.flowith.io/cn/flowith-normal/gao-ji-wan-fa-jie-dian-yin-yong#chun-wen-ben) 纯文本 **插入节点** 1. 点击左上角工具栏里 「变量节点」 按钮并拖拽到画布中 2. 修改节点的名字 3. 输入内容 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FPgiTvKeflNEubPlaMMIK%252F%25E6%258F%2592%25E5%2585%25A5%25E8%258A%2582%25E7%2582%25B9.gif%3Falt%3Dmedia%26token%3Da7541e82-a87c-41ec-9930-1372e09fb397&width=768&dpr=4&quality=100&sign=2e2766d0&sv=2) 在画布中插入一个节点 **引用节点** 1. 将光标移动到需要插入处 2. 点击画布下方的圆形按钮 3. 选择 「引用其他节点内容」 4. 点击你所要引用的节点 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FfdpEA9t9iwJYkV7hqWH5%252F%25E5%25BC%2595%25E7%2594%25A8%25E8%258A%2582%25E7%2582%25B9.gif%3Falt%3Dmedia%26token%3D5dca7966-d9f9-40d3-9de9-b28f5d6f9e8a&width=768&dpr=4&quality=100&sign=559c56c&sv=2) 当画布变暗时就可以选择要引用的节点 ### [](https://doc.flowith.io/cn/flowith-normal/gao-ji-wan-fa-jie-dian-yin-yong#shang-chuan-wen-ben-wen-jian-huo-tu-pian) 上传文本文件或图片 **上传文本文件** 1. 将光标移动到需要插入处 2. 点击画布下方的圆形按钮 3. 选择 「上传文本文件进行提问」 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FgeFu56O2i0GvZVg4UZzM%252F%25E4%25B8%258A%25E4%25BC%25A0%25E6%2596%2587%25E6%259C%25AC%25E6%2596%2587%25E4%25BB%25B6.gif%3Falt%3Dmedia%26token%3D811eaf6d-078b-4142-9a42-3387275b0979&width=768&dpr=4&quality=100&sign=618ea858&sv=2) 文本文件作为 「插入节点」 展现在画布中 **上传图片文件** 1. 将光标移动到需要插入处 2. 点击画布下方的圆形按钮 3. 选择 「上传图片进行提问」 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FJ8Ix2SlVtihhmpm3G7EA%252F%25E4%25B8%258A%25E4%25BC%25A0%25E5%259B%25BE%25E7%2589%2587.gif%3Falt%3Dmedia%26token%3Db2c7b982-c8b1-4c5e-a63a-8246cb623309&width=768&dpr=4&quality=100&sign=8a435c15&sv=2) 图片作为 「插入节点」 展现在画布中 [Previous流式聊天使用指南](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-shi-yong-zhi-nan) [NextPrompt Engineering 基础](https://doc.flowith.io/cn/flowith-normal/prompt-engineering-ji-chu) Last updated 11 months ago Was this helpful? --- # 知识花园功能简介 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FUopBhOuVucfMLtuovA6j%252F1280X1280.PNG%3Falt%3Dmedia%26token%3De7e4dc85-0590-4245-ada2-b4308f36ef2c&width=1248&dpr=4&quality=100&sign=16bcc41a&sv=2) [](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-gong-neng-jian-jie#wei-shen-me-yao-shi-yong-flowith-jin-xing-zhi-shi-guan-li) 为什么要使用 flowith 进行知识管理 ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FUopBhOuVucfMLtuovA6j%252F1280X1280.PNG%3Falt%3Dmedia%26token%3De7e4dc85-0590-4245-ada2-b4308f36ef2c&width=768&dpr=4&quality=100&sign=16bcc41a&sv=2) 不管你是内容创作者、学生、设计师、程序员、管理者还是任何形式的知识工作者，哪怕你已经对各类 AI 工具的使用了如指掌，你一定还是对我提到的这些痛点深有同感： 1. **信息离散无序**：笔记、文档分散在各个软件、文件夹、浏览器标签，知识无法形成体系化的沉淀，难以快速找到需要的信息；每次启动工作时，你面对的都是一个个空白页面，然后需要在无数文件、软件与页面间四处搜索和拷贝信息，重头开始或撰写或编撰成冗长的 prompt 喂给 AI。 2. **协作隔离低效**：你和你的团队成员的知识无法共享，每次都要寻找一个数个月甚至数年前的文档上的某段信息，导致沟通成本高昂、产出质量参差不齐。 3. **AI 生成内容没有针对性**：大语言模型过于通用，AI 缺乏个人知识和专业领域的理解，导致难以根据个人背景产生符合要求的内容并且容易产生"幻觉"给出错误信息，无法确保回答的一致性。 人们被迫屈从于格式和工具的桎梏，而非集中精力于知识本身的逻辑结构和创新表达，这在今天的 Generative AI 时代，显得格外落伍，犹如用打孔纸带来建造火箭。 [](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-gong-neng-jian-jie#ai-shi-dai-de-zhi-shi-guan-li-yu-chuang-zuo-shen-me-shi-flowith-zhi-shi-guan-li-mo-kuai) **AI 时代的知识管理与创作——**什么是 flowith 知识管理模块？ ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- 联通 AI 时代的知识管理和创作，flowith 2.0 推出了全新的 「Knowledge Garden」 知识花园，它是一个基于 AI Agent 的新一代知识管理系统，帮助你大幅提高 AI 创作效果与效率的同时进行自动化的个人知识管理： * **快速创建个人知识库**：无需从头开始，只要导入已有的资料和文件，flowith AI 就会借助 Seed 系统将其自动分解为知识的最小单元进行自动化知识管理、快速搭建你的 AI 知识库。 * **自动化匹配和生成：**在你使用 AI 进行特定任务时，flowith 会自动匹配你选择的知识库下有价值的信息，这让 flowith 知识库成为一个真正智能、有生命力的第二大脑系统，帮助你更自然地管理和运用知识。 * **联通知识管理和内容生成：**你的内容会被自动建立关联，形成一个有机且智能的知识网络，确保回答的准确性并有效避免“幻觉”现象，显著提升生成质量。 * **显著降低使用门槛：**专注在个人知识库的搭建，而非复杂的提示词工程上，即可与 AI 进行自然的对话，使用和共享社区知识库，帮助你大幅节省工作成本或是让你的知识盈利。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FgCaRsPaS3EQnyB6PT6y4%252F%25E7%2583%25AD%25E5%258A%259B%25E5%259B%25BE.PNG%3Falt%3Dmedia%26token%3D7ecf71dc-94fe-45a5-949d-b1def9f24025&width=768&dpr=4&quality=100&sign=c42b6bcf&sv=2) Knowledge Garden 功能将 LLM 的技术特性和人类卡片知识管理思维相结合，让 AI 更好的读懂、利用人类的私有化知识，帮助人们更有针对性、更高质量的生成内容。同时 Knowledge Garden 里的信息对人类来说，也可以用作更高效的知识管理。 总而言之，如果你的知识类型学习和工作是一场游戏，flowith 就为你打造了一个 「外挂」 级的 AI 知识管理系统，让你获得当前市面上所有产品中所能得到最高质量的 AI 生成效果。 [PreviousPrompt Engineering 基础](https://doc.flowith.io/cn/flowith-normal/prompt-engineering-ji-chu) [Next知识花园案例](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-an-li) Last updated 5 months ago Was this helpful? --- # Oracle 适合解决什么样问题 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FVXW8sFynBZ5itbEkvKq3%252FAmazing%2520Mockups.jpeg%3Falt%3Dmedia%26token%3Dcecb7610-ecef-4f75-8f0b-b6582b2c3813&width=1248&dpr=4&quality=100&sign=6927469e&sv=2) Oracle 模式适合解决多步骤、复杂、专业化的任务和需求。 在 Oracle 模式下，你可以提出一个宽泛的问题，AI 通过一步步的追问，使你的问题更加细化： ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FQV3vfctzNUt9k3qVfAb1%252F%25E5%25AD%25A6%25E4%25B9%25A0%25E6%2596%25B0%25E9%25A2%2586%25E5%259F%259F.jpeg%3Falt%3Dmedia%26token%3Dafc11afd-65fb-4159-b3c5-a7482b947c97&width=768&dpr=4&quality=100&sign=69d89333&sv=2) 信息不足时 Oracle 进行追问 不仅如此，Oracle 模式可以根据你的问题，自主规划、拆分并完成复杂任务，并且同时生成一个具体可行的执行步骤。从而更加适合多步骤且复杂的问题。 如果你有调研一个产品或现象并做成报告的需求，则 Oracle 会非常合适，因为它不仅可以更全方位的在各个平台进行搜索，也可以利用它生成网页、PPT 的能力，为你更好的呈现你想要的信息。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252Fuy0KfP4kvVJBeU7wbBhx%252Fmnggiflab-compressed-Webnew.gif%3Falt%3Dmedia%26token%3D4cbf1aae-4f42-4e9c-9913-4d5cae897e09&width=768&dpr=4&quality=100&sign=46802671&sv=2) Oracle 生成的网页 除此之外，你还可以在社区中搜索、查看或使用专业人士给出的相关模板，可以帮助你更加轻松且清晰地解决你提出的问题。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FSc2Y9suzZ3nzmpsdDvmp%252F%25E7%25A4%25BE%25E5%258C%25BA%2520%281%29.jpeg%3Falt%3Dmedia%26token%3D30c5ac30-44e9-43ee-b157-e39a63404ba5&width=768&dpr=4&quality=100&sign=7aae3e8f&sv=2) 社区中有各种专业模板可供使用 [PreviousOracle 目前所支持的能力](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mu-qian-suo-zhi-chi-de-neng-li) [NextOracle 模式使用指南](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mo-shi-shi-yong-zhi-nan) Last updated 10 months ago Was this helpful? --- # Oracle 目前所支持的能力 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FVXW8sFynBZ5itbEkvKq3%252FAmazing%2520Mockups.jpeg%3Falt%3Dmedia%26token%3Dcecb7610-ecef-4f75-8f0b-b6582b2c3813&width=1248&dpr=4&quality=100&sign=6927469e&sv=2) 以下为 Flowith Oracle 模式目前支持的外部工具，这个列表正在持续更新中，如果你有一些需要的工具希望 Oracle 可以支持，或希望将自己产品的 API 和 Oracle 相结合，欢迎和我们联系！ Tool 工具名称 工具作用 analyze\_website\_traffic analyze specific website traffic; get website metrics; check website engagement ask\_user\_more ask user to input more information deep\_thinking deep analysis on a subject gen\_webpage generate web page, create website, design user interface, build online platform, develop web application, construct digital presence, craft internet portal, produce web-based solution, create ppt presentations, design infographics, generate charts and graphs, visualize data, create mindmaps and flowcharts. generate\_image generate images; draw images; text to image generations; draw a cute cat generate\_python help user run python code,code interpreter; do complex mathematical operations; do data visualization; run python code online\_search search realtime information by Google read\_website read a website's content; scrape website information; retrieve online articles search\_google\_scholar search academic papers; find scholarly articles; search\_images google image search; search\_jobs search for jobs search\_linkedin\_company search company on LinkedIn search\_linkedin\_people search people on LinkedIn, find LinkedIn profiles search\_patents search patents, find patent information, lookup inventions search\_products search for products, find items to buy, compare prices, shopping information search\_reddit\_posts search Reddit posts search\_stackexchange\_question search stackoverflow posts search\_twitter search tweets, find twitter posts search\_videos search YouTube videos, video search search\_websites search for websites, find links, google search search\_xiaohongshu\_notes search xiaohongshu posts send\_email send email, send mail, g-mail, contact a person by e-mail send\_rapid\_email send email; fast email delivery summarize\_and\_decide summarize contents translate\_text translate content user provided [Previous关于 Oracle 模式](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/guan-yu-oracle-mo-shi) [NextOracle 适合解决什么样问题](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-shi-he-jie-jue-shen-me-yang-wen-ti) Last updated 11 months ago Was this helpful? --- # 知识花园案例 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) 以下是我们产品团队在内测和公测期间收集的一些令人印象深刻的案例： 1. **媒体内容创作（Content Creation）——社区案例** Alex 是一位小红书内容创作者，他的笔记以高质量内容吸引了众多粉丝。然而由于纯原创的内容跟不上最新的热点他的增长陷入了瓶颈。在我们的内测期间，他将之前的写作素材和过往文章上传到了 flowith knowledge base。在撰写新文案时，flowith 自动帮他匹配相关素材，并依据他的文风实现了快速的内容和个性化生成。比起之前每次都需要重头开始写作，现在的他只需要在输入节点内更换一下新的写作题材就可以快速生成符合自己文章风格、能运用个人素材的高质量文案，据他称帮他“减少了80% 的时间”——现在他只需要专注在构思有趣的点子就可以了。 1. **学生进行考试备考、文书写作（Education&Academia）——公测用户案例** Ze 正准备她的课程期末，她将某门课程的课件和参考资料导入 flowith 知识库，系统在她撰写时即时提供所需知识点。Ze 发现，借助这些功能，她不仅论文的写作效率提高，同时备考过程也轻松愉快，再也不用苦苦翻找资料，学习变得更智能高效。 1. **基于知识库进行团队协作（Team Collaboration）——公测用户案例** 某教育科技公司的市场部门每年需要面向不同用户群体产出上百篇营销内容。在采用 flowith 前，团队的素材分散在各个平台，查找和复用效率低下。现在他们将所有市场资料统一导入 flowith 知识库，AI 能自动调用相关素材并生成符合团队风格的内容。这不仅极大降低了重复劳动，更重要的是让团队产出的质量和风格更加统一。 1. **学术研究（Research）——公测用户案例** 内测用户某高校教授使用 flowith 管理他学术生涯收集的资料，自动汇聚不同来源的数据与文献支持他的研究，为学术贡献带来更广泛的视野和更深刻的洞见。 1. **教师备课——公测用户案例** 某高校教师面临一个挑战：需要将过去 5 年来积累的上百篇论文、教案和课件整合，准备新学期的系列课程。在导入 flowith 后，AI 自动将这些资料分解为知识种子(Seeds)并建立关联。现在他只需输入课程主题，flowith 就能智能匹配相关内容，让备课过程事半功倍。"这让我的教学准备时间减少了 60%，课程质量反而更高了。"该教师如是说。 1. **德州扑克——社区案例** 社区用户 Xavier 在内测群内发布了德州扑克的知识库，他收集了职业选手的比赛分析、策略文章和实战案例，获得了一个可以帮助学习和动态帮助在德州扑克游戏中进行高质量决策的“外挂”。可以看出，它的效果比普通 AI 给出的要好得多。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252F5P2dfJ60u33kmVz9vuRT%252F%25E5%25BE%25B7%25E5%25B7%259E%25E6%2589%2591%25E5%2585%258B.png%3Falt%3Dmedia%26token%3D7234b874-f65a-4689-91a5-2a90162dda62&width=768&dpr=4&quality=100&sign=a197971c&sv=2) 1. **大学申请——社区案例** 一位教育顾问将数百份成功申请 Harvard、MIT、Yale 等名校的文书范例整理成知识库上线到了 flowith Marketplace。这个知识库不仅能分析这些成功案例的共同特征，还能根据申请者的背景量身定制文书框架、产出高质量的初始文案。该知识库上线一周内就创造了超过 800 美元的收入。 总而言之，只要是能够通过个人信息和知识增强提升 AI 生成效果的领域，flowith 知识库几乎适用于任何类型的知识工作，并带给你当前 AI 产品中最佳的创造效果。 [Previous知识花园功能简介](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-gong-neng-jian-jie) [Next如何建立和使用知识库](https://doc.flowith.io/cn/knowledge-garden/ru-he-jian-li-he-shi-yong-zhi-shi-ku) Last updated 7 months ago Was this helpful? --- # 如何建立和使用知识库 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) [](https://doc.flowith.io/cn/knowledge-garden/ru-he-jian-li-he-shi-yong-zhi-shi-ku#jian-li-zhi-shi-ku) 建立知识库 ----------------------------------------------------------------------------------------------------------------- 1. 点击 「管理你的知识库」 2. 点击左上角的加号，添加新的知识库 3. 添加文件 1. 文件支持的格式：image、PDF、markdown、txt、ms-officie 2. 也可通过网页链接进行导入 [](https://doc.flowith.io/cn/knowledge-garden/ru-he-jian-li-he-shi-yong-zhi-shi-ku#shi-yong-zhi-shi-ku) 使用知识库 ------------------------------------------------------------------------------------------------------------------ 1. 打开输入框右下角的 「KNOWLEDGE BASE」 ，并勾选你需要的知识库 2. 选择模式、模型 3. 在输入框中输入问题 4. 点击开始进行提问 [](https://doc.flowith.io/cn/knowledge-garden/ru-he-jian-li-he-shi-yong-zhi-shi-ku#shi-yong-she-qu-zhong-de-zhi-shi-ku) 使用社区中的知识库 -------------------------------------------------------------------------------------------------------------------------------------- 1. 点击页面左上角 「发现更多」 2. 点击你感兴趣的知识库 3. 点击页面右方 「添加到我的知识库 」 4. 按照上述 「使用知识库」 中的步骤进行使用 [Previous知识花园案例](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-an-li) [Next知识花园变现指南](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-bian-xian-zhi-nan) Last updated 7 months ago Was this helpful? --- # 关于 Oracle 模式 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FdhFjmMPclhNzJp7dhZ1c%252F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png%3Falt%3Dmedia%26token%3D02f9f800-a62e-4b49-9d4b-20327cfdece0&width=1248&dpr=4&quality=100&sign=7c56180c&sv=2) [](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/guan-yu-oracle-mo-shi#shen-me-shi-oracle) 什么是 Oracle --------------------------------------------------------------------------------------------------------------------- 这是由 flowith 团队开发的最先进的、以通用人工智能（pre-AGI）为目标设计的的智能体系统，专为处理多步骤、复杂、需要使用更多工具的任务而设计。 它的功能包括自主规划任务、任务分解、无限制工具使用和自我优化。Oracle 的外部能力会不断更新和升级。使其随着时间的推移变得越来越强。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FCIDy6htAMfmo3KDXw8ZE%252FAmazing%2520Mockups.jpeg%3Falt%3Dmedia%26token%3Db5dd6082-4b4d-4f4a-978a-c5fc8a9bc77e&width=768&dpr=4&quality=100&sign=8ac5926a&sv=2) Oracle Recipe and Nodes **换句话说，Oracle is a data-level**（数据层级别的、而不具备物理介质或和现实世界交互的）generalized (可以完成通用的任务、而非只有他被训练、或定义过的任务) **Agent**（不只是 LLM，可以调用工具、管理记忆、多 LLM 协作） that can complete complex tasks （不只是完成简单的问题或命令）and adapt to new situations（在新的环境或任务下可以自我学习和处理）. [](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/guan-yu-oracle-mo-shi#he-xin-gai-nian) 核心概念 ------------------------------------------------------------------------------------------------------------ **Recipe 工作流** Recipe 一词原意是指配方、烹饪方法。对于 Oracle 来说，Recipe 是指 AI 在去完成用户任务时，去执行的一套系统性工作流。当每次用户输入完需求或指令，Oracle 都会整体的指令进行拆解，根据它自身的能力和可调用的工具，把整体任务拆成一步一步自任务，然后再交给任务分配器去一步步执行。 在每次 Oracle 设计完一整套 Recipe 后，它都会与用户确认，随后才开始执行，用户在确认过程中可以直接修改 Recipe 里的具体执行步骤，包括调整顺序、删除、修改文字和增加新的步骤。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FEFYpRO4ie9zxqLfGqxad%252F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png%3Falt%3Dmedia%26token%3Deef0fe07-00ce-40fe-af4d-b8d5209b5823&width=768&dpr=4&quality=100&sign=7db10423&sv=2) A Recipe Generated by Oracle **Dispatcher 任务分配器** 任务分配器是指每一条 Oracle 生成的自任务在被执行时，会有一个分配任务的 Agent 把子任务进行分配和落地执行，它会从 Oracle 支持的工具库里找到最合适的工具，然后把子任务进行执行和完成。 比如，当子任务是 「搜索 flowith 在社交媒体上的用户评价」 ，任务分配器会调用 「Twitter 搜索工具」 、 「Reddit 搜索工具」 、「谷歌搜索工具」 进行对 flowith 用户评价的搜索。 你可以在 Oracle 对话发起页面中查看目前 Oracle 所支持的工具，这个列表会实时更新，我们也将持续把最有用的工具添加到 Oracle 所支持的工具库中。 [Oracle 目前所支持的能力](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mu-qian-suo-zhi-chi-de-neng-li) ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FTMrTvG0vDQ7aeyQ9mim6%252F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE4.jpg%3Falt%3Dmedia%26token%3D7a3791e7-d36f-48f0-bc6c-e65c5203315f&width=768&dpr=4&quality=100&sign=3213cf5b&sv=2) Tools supported by Oracle **Memory 记忆和上下文管理** Oracle 是为复杂工作而生，因此对于步骤较多的工作，会产生大量的上下文，如果我们将所有上文内容交给 LLM，则将影响它的工作表现。 因此 Oracle 在任务分配、执行的过程中，会智能的选择性使用上文信息，比如当 Oracle 执行到第 10 步时，它的记忆管理器会判断以上哪些步骤的信息是有必要的，并将必要的信息放入当前步骤的上文。因此，你无需担心上文过长的问题。 **自我迭代和更新 Recipe** 在每一步执行完后，Oracle 都会根据每一步骤的执行结果，对整个 Recipe 进行必要的更新，比如当它没有搜到需要的信息，它会给接下来的步骤中加入更多的搜索，以便它可以顺利完成任务。 当 Oracle 询问是否满意结果时，如果用户表示不满意，它也将生成更多的步骤，尝试为用户提供想要的结果。 [Previous知识花园变现指南](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-bian-xian-zhi-nan) [NextOracle 目前所支持的能力](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mu-qian-suo-zhi-chi-de-neng-li) Last updated 10 months ago Was this helpful? --- # Oracle 模式高级玩法 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FdhFjmMPclhNzJp7dhZ1c%252F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png%3Falt%3Dmedia%26token%3D02f9f800-a62e-4b49-9d4b-20327cfdece0&width=1248&dpr=4&quality=100&sign=7c56180c&sv=2) [](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mo-shi-gao-ji-wan-fa#ru-he-xiu-gai-recipe) 如何修改 Recipe ------------------------------------------------------------------------------------------------------------------------------ * 当一条 recipe 太长时，可以在触控板上滑动来读取后面的内容 * 需要删除某条 recipe 时，勾选该条后方的 「垃圾桶」 图标 * 点击页面左下角的 「＋」 ，即可添加一条新的步骤 * 点击某条前方的 「拖拽」 符号并拖拽，即可把该条放在你想要的位置 * 调整完全部 recipe 后，点击 「Confirm & Run」 ，即可运行 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FSN0p5uYzykl7QJ0ZpFut%252F%25E4%25BF%25AE%25E6%2594%25B9recipe.gif%3Falt%3Dmedia%26token%3Df7d2928a-d308-4957-9684-f1fa38bf8f44&width=768&dpr=4&quality=100&sign=eeec7f93&sv=2) 修改 recipe [](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mo-shi-gao-ji-wan-fa#ru-he-jiang-zi-ji-zhi-ding-de-recipe-fen-xiang-dao-she-qu) 如何将自己制定的 Recipe 分享到社区 ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------- 你可以将你你反复打磨的 recipe 分享到社区，让大家一起来体验你的智慧成果。 1. 点击 「Share」 2. 点击 「Publish to Community」 3. 填写必要信息并发布！ ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FXJcXJIUwNcrva5MROirO%252F%25E5%2588%2586%25E4%25BA%25ABrecipe.gif%3Falt%3Dmedia%26token%3D2e9cd98e-22c7-45d5-8616-f90f4191eff1&width=768&dpr=4&quality=100&sign=9c90255b&sv=2) 详细填写模板信息 [PreviousOracle 模式使用指南](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mo-shi-shi-yong-zhi-nan) [Next关于订阅](https://doc.flowith.io/cn/qi-ta-wen-ti/guan-yu-ding-yue) Last updated 11 months ago Was this helpful? --- # flowith ナレッジガーデンの概要 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdenno#naze-flowith-dewounoka) なぜ flowith で知識管理を行うのか ------------------------------------------------------------------------------------------------------------------------------------------------- ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F2916067171-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252F2agAjJERs7Y6IiotYFD6%252Fuploads%252FJT059p2sWZQHj73T8apj%252F1280X1280.PNG%3Falt%3Dmedia%26token%3Dc70c21f3-9953-4ac0-b910-a31d89a37db5&width=768&dpr=4&quality=100&sign=16487de0&sv=2) コンテンツクリエイター、学生、デザイナー、プログラマー、マネージャー、その他どのような形態のナレッジワーカーであっても、また、様々な AI ツールの使用に精通されている方でも、以下のような課題に共感されることと存じます： 1. **情報の散在と無秩序**：ノート、文書が様々なソフトウェア、フォルダ、ブラウザのタブに分散しており、知識が体系的に蓄積されず、必要な情報を素早く見つけることが困難です。作業を開始するたびに、空白のページから始まり、無数のファイル、ソフトウェア、ページの間を行き来して情報を検索・コピーし、最初から作成するか、AI に与える長いプロンプトを作成する必要があります。 2. **協働の分断と非効率**：チームメンバーとの知識共有ができず、数ヶ月前や数年前の文書から特定の情報を探さなければならず、コミュニケーションコストが高騰し、成果物の品質にばらつきが生じています。 3. **AI生成コンテンツの的確性の欠如**：大規模言語モデルが一般的すぎるため、AI は個人の知識や専門分野への理解が不足しており、個人の背景に基づいた要件に合うコンテンツを生成することが困難です。また、「 幻覚 」 により誤った情報を提供しやすく、回答の一貫性を確保できません。 人々は、知識それ自体の論理構造や革新的な表現に集中するのではなく、フォーマットやツールの制約に従わざるを得ない状況に置かれています。これは現代の Generative AI 時代において、特に時代遅れに見え、まるで紙テープでロケットを作ろうとするようなものです。 [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdenno#ai-noniokeruto-flowith-nomojrutoha) AI の時代における知識管理と創造 ー flowith の知識管理モジュールとは -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- AI時代のナレッジマネジメントとコンテンツ創作をつなぐ、flowith 2.0は新しい「Knowledge Garden」ナレッジガーデンを導入しました。これはAI Agentベースの次世代ナレッジマネジメントシステムであり、AI創作の効果と効率を大幅に向上させながら、個人の知識を自動的に管理します： * **個人ナレッジベースの迅速な構築**：ゼロからの開始は不要です。既存の資料やファイルをインポートするだけで、flowith AI はSeedシステムを活用し、それらを知識の最小単位に自動的に分解し、自動的なナレッジマネジメントを行い、AI ナレッジベースを迅速に構築します。 * **自動マッチングと生成**：特定のタスクで AI を使用する際、flowith は選択したナレッジベースから価値のある情報を自動的にマッチングします。これにより、flowith のナレッジベースは真に知的で生命力のあるセカンドブレインシステムとなり、より自然な知識の管理と活用を支援します。 * **ナレッジマネジメントとコンテンツ生成の統合**：コンテンツは自動的に関連付けられ、有機的かつインテリジェントな知識ネットワークを形成し、回答の正確性を確保し、「 幻覚 」 現象を効果的に防ぎ、生成品質を大幅に向上させます。 * **使用の敷居を大幅に下げる**：複雑なプロンプトエンジニアリングではなく、個人ナレッジベースの構築に注力するだけで、AIと自然な対話が可能になり、コミュニティナレッジベースの使用と共有により、作業コストを大幅に削減したり、知識を収益化したりすることができます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F2916067171-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252F2agAjJERs7Y6IiotYFD6%252Fuploads%252FvO19HWCpQK3XaFk5Pf9L%252F%25E7%2583%25AD%25E5%258A%259B%25E5%259B%25BEn.PNG%3Falt%3Dmedia%26token%3D71d0c96b-cf31-47f7-8677-1172373b11bf&width=768&dpr=4&quality=100&sign=c8331c5&sv=2) Knowledge Garden は、LLM の技術的特性と人間のカード型知識管理思考を組み合わせることで、AI がより効果的に人間の個人的な知識を理解・活用し、より的確で質の高いコンテンツ生成をサポートする機能です。また、Knowledge Garden 内の情報は、人間にとってもより効率的な知識管理のツールとして活用できます。 つまり、あなたの知識習得や仕事がゲームだとすれば、flowithは「チートツール」のような AI ナレッジマネジメントシステムを提供し、現在市場に存在する全製品の中で最高品質の AI 生成結果を実現します。 [Previousプロンプトエンジニアリングの基礎](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/puronputoenjiniaringuno) [Nextflowith ナレッジベース・ユースケース](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjibsuysuksu) Last updated 7 months ago Was this helpful? --- # ナレッジベースの構築と使用方法 | Flowith Official Document [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/narejjibsunoto#narejjibsuno) ナレッジベースの作成 --------------------------------------------------------------------------------------------------------------------- 1. 「 ナレッジベース管理 」 をクリック 2. 左上のプラス記号をクリックして、新規ナレッジベースを追加 3. ファイルの追加 1. 対応ファイル形式：画像、PDF、markdown、txt、ms-office 2. Web ページリンクからのインポートも可能 [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/narejjibsunoto#narejjibsuno-1) ナレッジベースの使用方法 ------------------------------------------------------------------------------------------------------------------------- 1. 入力欄右下の KNOWLEDGE BASE を開き、必要なナレッジベースにチェックを入れます 2. モードとモデルを選択します 3. 入力欄に質問を入力します 4. 「 開始 」 をクリックして質問を実行します [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/narejjibsunoto#narejjibsuno-2) ナレッジベースの利用方法 ------------------------------------------------------------------------------------------------------------------------- 1. 画面左上の 「 もっと見る 」 をクリックします 2. 興味のあるナレッジベースをクリックします 3. 画面右側の 「 マイナレッジベースに追加 」 をクリックします 4. 上記の 「 ナレッジベースの使用方法 」 の手順に従って操作します [Previousflowith ナレッジベース・ユースケース](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjibsuysuksu) [Nextflowith ナレッジガーデン収益化ガイド](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdengaido) Last updated 7 months ago Was this helpful? --- # 关于订阅 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) [](https://doc.flowith.io/cn/qi-ta-wen-ti/guan-yu-ding-yue#ding-yue-fang-an) 订阅方案 -------------------------------------------------------------------------------------- 我们提供了极具性价比的订阅方案，提供 OpenAI、Anthropic、Google、DeepSeek、Meta 等公司的顶尖 AI 模型支持。除此之外，我们也接入了支持生成图片和视频的顶尖模型。 详情请见：[https://flowith.io/sub](https://flowith.io/sub) [](https://doc.flowith.io/cn/qi-ta-wen-ti/guan-yu-ding-yue#chao-ji-mo-xing-oracle) 超级模型—— Oracle ----------------------------------------------------------------------------------------------------- 在 Oracle 模式中，每一步都会消耗一个 T-1 配额。例如，一个需要六个步骤的任务将使用 6 个 T-1 额度。 [](https://doc.flowith.io/cn/qi-ta-wen-ti/guan-yu-ding-yue#zhi-shi-ku) 知识库 ------------------------------------------------------------------------------- 1. **文件存储计费机制** 文件上传至知识库时，仅计入知识库存储空间配额，不产生额外的模型调用费用。 1. **对话服务计费规则** 当通过知识库发起对话时，系统将根据所选用模型的服务层级（T1级别、T2级别）进行模型调用次数的计量计费，计费标准与基础对话服务保持一致。 1. **存储配额管理** 关于知识库存储配额的具体限制及管理规范，请参阅：[https://flowith.io/sub](https://flowith.io/sub) [](https://doc.flowith.io/cn/qi-ta-wen-ti/guan-yu-ding-yue#zhu-yi-shi-xiang) 注意事项： --------------------------------------------------------------------------------------- **政策说明：** 我们发现部分用户通过批量注册大量 Flowith 账号，滥用免费额度调用我们成本高昂的 AI 模型，这种行为严重影响了其他用户的正常使用体验。 若您遇到特殊情况达到调用上限，请联系我们。 联系方式： 邮箱：[\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection) Discord：[https://discord.com/invite/g95B77xbpD](https://discord.com/invite/g95B77xbpD) ### [](https://doc.flowith.io/cn/qi-ta-wen-ti/guan-yu-ding-yue#tui-kuan-zheng-ce) 退款政策 * 订阅费用通常不可退款 * 仅在同时满足以下情况可申请退款： * 在支付成功后48小时内联系我们 * 订阅后账户累计使用Credits少于200 如遇以下情况，请联系 [\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection) * 任何未尽事宜或特殊情况 * 对于符合退款条件的申请 * 关于取消订阅的技术问题咨询 [PreviousOracle 模式高级玩法](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mo-shi-gao-ji-wan-fa) [Next如何反馈问题](https://doc.flowith.io/cn/qi-ta-wen-ti/ru-he-fan-kui-wen-ti) Last updated 2 months ago Was this helpful? --- # 如何反馈问题 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) 如果您在使用过程中遇到任何问题或者有任何建议，欢迎联系我们。 **通过邮箱** 官方邮箱：[\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection) **通过产品页面** 1. 点击产品页面 「[flowith.io](https://flowith.io/) 」 的左下角 「订阅与设置」 2. 点击 「订阅与设置」 中的 「反馈联系」 3. 加入discord [Previous关于订阅](https://doc.flowith.io/cn/qi-ta-wen-ti/guan-yu-ding-yue) [Next信息搜集 + 整理大师](https://doc.flowith.io/cn/ying-yong-an-li/xin-xi-sou-ji-+-zheng-li-da-shi) Last updated 6 months ago Was this helpful? --- # Oracle モードの高度なテクニック | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fzu8C1nUY5LDCaMxYYNnm%2F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png&width=1248&dpr=4&quality=100&sign=4dfcd71&sv=2) [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdononatekunikku#recipe-no) **Recipe の修正方法** ------------------------------------------------------------------------------------------------------------------------- * recipe が長すぎる場合は、タッチパッドをスワイプして残りの内容を読むことができます。 * 特定の recipe を削除するには、その横にある 「ゴミ箱」 アイコンを選択してください。 * ページの左下にある「+」ボタンをクリックすると、新しいステップを追加できます。 * ステップの順序を変更するには、ステップの前にある 「ハンドル」 アイコンをクリックしてドラッグし、希望の位置に移動させてください。 * すべての recipe の調整が完了したら、「確認して実行」 をクリックして実行します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FGi0ihgRagg5cddLoGsap%2FOracle-%25E9%25AB%2598%25E7%25BA%25A7%25E7%258E%25A9%25E6%25B3%2595-%25E4%25BF%25AE%25E6%2594%25B9recipe.gif&width=768&dpr=4&quality=100&sign=6bf1f0bc&sv=2) recipe の修正 [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdononatekunikku#komyunitinikasutamu-recipe-wosuru) **コミュニティにカスタム recipe を共有する方法** --------------------------------------------------------------------------------------------------------------------------------------------------------------- あなたが丁寧に作成したレシピをコミュニティと共有することで、他のユーザーもあなたの創造的なアイデアの恩恵を受けることができます。 recipe を共有するには： 1. 「Planner」 セクションに移動します。 2. 「Share to Community」 ボタンをクリックします。 3. 必要な情報を入力します。 4. recipe を公開します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FIMTus75mgDj2NtdVNw83%2FOracle-%25E9%25AB%2598%25E7%25BA%25A7%25E7%258E%25A9%25E6%25B3%2595-%25E5%258F%2591%25E5%25B8%2583%25E5%2588%25B0%25E7%25A4%25BE%25E5%258C%25BA.gif&width=768&dpr=4&quality=100&sign=5b65e707&sv=2) テンプレートの詳細情報 [PreviousOracle モード ユーザーガイド](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdo-yzgaido) [Nextご契約に関して](https://doc.flowith.io/flowith-heyoukoso/sononogo/gonishite) Last updated 10 months ago Was this helpful? --- # 信息搜集 + 整理大师 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) You can now have dozens, even hundreds of AI Agents working for you simultaneously. Oracle is a more practical and powerful agent framework than AutoGPT. It autonomously plans tasks, designs workflows, calls tools, and iterates on subsequent workflows. [Previous如何反馈问题](https://doc.flowith.io/cn/qi-ta-wen-ti/ru-he-fan-kui-wen-ti) Last updated 11 months ago Was this helpful? --- # Flowith へようこそ | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FzmBhbFHFJ6G9rBN4Fscb%2FOriginal5.jpg&width=1248&dpr=4&quality=100&sign=395fc73b&sv=2) flowith - AI for Deep Work [](https://doc.flowith.io/flowith-heyoukoso#flowith-toha) **flowith とは？** ------------------------------------------------------------------------------ flowith は、次世代の AI 生産性ツールです。従来のチャットベースの AI ツールとは異なり、flowith は二次元キャンバスを基にした対話方式を採用しています。これにより、単一のワークスペース内で複数のAI モデルと同時に対話することが可能となり、効率を大幅に向上させるコンテンツ生成や、創造プロセスにおけるフロー状態への没入を促進します。 flowith Oracle（フロウィズ オラクル）は、当社独自の次世代 AI エージェントシステムで、複数のステップを要する高度に複雑なタスクや要件を遂行するのに役立つよう設計されています。以下の機能をサポートしています： * 自律的な計画立案 * タスクの分解と完了 * 無制限のツール呼び出し * 自己最適化と改善 **Oracle** は、AutoGPT などの従来のアーキテクチャと比較して、より成熟した安定したパフォーマンスを発揮します。さらに、flowithのキャンバスベースのインタラクションを活用することで、Oracle はより具体的な手順や包括的な情報をユーザーの皆様に分かりやすく提示することができます。 flowith **「Knowledge Garden」** ナレッジガーデンは、「AIとあなたの知識を自然に連携させる」 という中核理念に基づいて開発されました。自動的な知識整理、構造化された管理、そして創造的な協働を統合しています。文書、ノート、オンラインリソースを自動的にAIが理解できる最小知識単位（Seeds：ナレッジシード）に変換することができます。システムはAIによる知的分析を通じて、知識間の関連性を自動的に確立し、動的に成長する知識ネットワークを形成します。AIとの対話において、関連する知識が自動的に呼び出され、AIの応答の正確性と的確性が大幅に向上します。 製品情報: [flowith.io](https://flowith.io/) 公式ウェブサイト: [try.flowith.io](https://try.flowith.io/) 公式Twitter: @flowith\_ai お問い合わせ先: [\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection) Discord: [https://discord.com/invite/g95B77xbpD](https://discord.com/invite/g95B77xbpD) [](https://doc.flowith.io/flowith-heyoukoso#naze-flowith-wounoka) なぜ flowith を使うのか？ ---------------------------------------------------------------------------------------- **より効率的なインタラクション** 従来のチャットUI の直線的な対話方式と比較して、flowith は二次元キャンバス形式のインタラクションを採用しています。これにより、以下のことが可能になります： * 複数のスレッドで会話を展開 * 複数の分岐を作成 * 以下のようなシナリオに適しています： * 長文コンテンツの生成 * AI生成結果の比較 * プロンプトのデバッグ これらの機能により、flowith はさまざまな複雑なタスクや創造的なプロセスに柔軟に対応できます。 **豊富なモデル選択** 単一の大規模言語モデルに限定されたやり取りとは異なり、当プラットフォームでは、ユーザーが選択できる多様なモデルを提供しています。これには以下が含まれます： * OpenAI: o1 Preview、GPT-4o、4o-mini、Claude 3.5 sonnet などのマルチモーダルモデル * Stable Diffusion や DALL·E 3 などの画像生成モデル この多様性により、ユーザーは特定のニーズや用途に最適なモデルを選択することが可能になります。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FTtVJTQsDm4u5uvs2BZh2%2F%25E6%25AC%25A2%25E8%25BF%258E-%25E4%25B8%25B0%25E5%25AF%258C%25E7%259A%2584%25E6%25A8%25A1%25E5%259E%258B%25E9%2580%2589%25E6%258B%25A9.png&width=768&dpr=4&quality=100&sign=10d7fc3a&sv=2) **リアルタイムコラボレーションと双方向機能** flowith のキャンバスベースのインターフェースは、複数ユーザーによる共同作業に本質的に適しています。フローのリンクを同僚と共有することで、以下のことが可能になります： * 生成された結果を直接閲覧する * あなたの創作プロセスに参加する * 協力してコンテンツを生成し、デバッグする ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2F3qgCSWIVyADQFHDGk9ZE%2F%25E6%25AC%25A2%25E8%25BF%258E-%25E6%259B%25B4%25E5%25A4%259A%25E7%259A%2584%25E4%25BA%2592%25E5%258A%25A8%25E5%25BD%25A2%25E5%25BC%258F-%25E5%2588%2586%25E4%25BA%25AB.gif&width=768&dpr=4&quality=100&sign=66ae3155&sv=2) 友達とリンクを共有 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2F7SZetZqxjiDUJYzX88hU%2F%25E6%25AC%25A2%25E8%25BF%258E-%25E6%259B%25B4%25E5%25A4%259A%25E7%259A%2584%25E4%25BA%2592%25E5%258A%25A8%25E5%25BD%25A2%25E5%25BC%258F-%25E8%25AF%2584%25E8%25AE%25BA.gif&width=768&dpr=4&quality=100&sign=b202f684&sv=2) Eコメント機能で対話的なディスカッションに参加 **効率的な問い合わせ方法** Oracle モードでは、手動でのプロンプトエンジニアリングが不要になります。ユーザーは単に質問や要件を入力するだけで、Oracle が自律的に複雑なタスクの計画、分解、完了を行います。自動的に実行可能で編集可能な具体的なタスクステップを生成し提示するため、より実現可能で実用的な結果が得られます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fd7ts3tVBaGoBgZE3dlAB%2F%25E6%25AC%25A2%25E8%25BF%258E-%25E4%25BE%25BF%25E6%258D%25B7%25E7%259A%2584%25E6%258F%2590%25E9%2597%25AE%25E6%2596%25B9%25E5%25BC%258F.gif&width=768&dpr=4&quality=100&sign=9d49b80b&sv=2) オラクルモード：自動タスク分解 **ガイド付きフォローアップ質問** 従来の AI ツールは、ユーザーに必要な情報を尋ねることができず、曖昧で一般的な結果しか生成できないことがありました。一方、Oracle モードは積極的なアプローチを取ります。このモードでは、段階的な対話を行い、フォローアップ質問を通じて包括的かつ必要な詳細情報を収集します。この過程を経ることで、最終的にユーザーの具体的なニーズに合わせた高品質で深みのある最終出力を生成することができます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fl81iVMczv8ILLFGxGDj5%2F%25E6%25AC%25A2%25E8%25BF%258E%2520-%2520%25E5%25BC%2595%25E5%25AF%25BC%25E5%25BC%258F%25E7%259A%2584%25E8%25BF%25BD%25E9%2597%25AE.gif&width=768&dpr=4&quality=100&sign=1214a1bb&sv=2) フォローアップ質問で情報を充実 **無制限のツール連携** Oracle モードは、外部ツール（プラグインと考えることができます）との連携をサポートしています。これには、Twitter やシャオホンシュー（小紅書）の検索、要約や推論、マインドマップの生成、PowerPoint プレゼンテーションの作成、ウェブページの生成などが含まれます。Oracle は、現在のサブタスクに応じて、ツールライブラリから最適なツールを自動的に選択し、必要に応じて呼び出します。 Oracle がサポートするツールの数は継続的に増加しており、そのフレームワークは無制限のツール連携を可能にしています。これにより、ユーザーは多様な機能を柔軟に活用することができます。 * * * 私たちは、人間同士のやりとりや協力と同様に、人間と AI の最適な対話方法も発散的、マルチスレッド的、そして協調的であるべきだと考えています。そのため、従来の会話型 GPT アシスタントとは全く異なる、革新的な AI 対話手法を開発しました。無限のキャンバス上で AI と自由に協働することで、アイデア創出の可能性を無限に広げると同時に、明確なプロセスを通じてアイデアを再利用可能なワークフローに変換することができます。 さあ、Flo を使って始めましょう！ [](https://doc.flowith.io/flowith-heyoukoso#kuikkusutto) **クイックスタート** -------------------------------------------------------------------------- [Flow モード ユーザーガイド](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdo-yzgaido) [NextFlow モードの紹介](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdono) Last updated 2 months ago Was this helpful? --- # Oracle モードはどのような問題解決に適していますか？ | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FLb1TccN0scGPrEBlzdlD%2FAmazing%2520Mockups.jpeg&width=1248&dpr=4&quality=100&sign=50a313dc&sv=2) Oracle モードは、複数のステップを要する複雑な課題や専門的な要件に対応するのに最適です。 Oracle モードでは、幅広い質問を提示すると、AI が一連の追加質問を通じて問題を絞り込み、具体化していきます： ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FyNbW5bohrczwWLOHQrIc%2FOracle-%25E4%25BB%2580%25E4%25B9%2588%25E6%25A0%25B7%25E7%259A%2584%25E9%2597%25AE%25E9%25A2%2598-1.%25E7%25BB%2586%25E5%258C%2596%25E9%2597%25AE%25E9%25A2%2598.png&width=768&dpr=4&quality=100&sign=561a9582&sv=2) 情報不足時の Oracle モードの追加質問機能 さらに、Oracle モードは、ユーザーの質問に基づいて独自に計画を立て、複雑なタスクを分解し、実行することができます。同時に、具体的で実行可能な実行プランを生成します。これにより、複数のステップを含む複雑な問題に特に適しています。 製品や現象について調査し、レポートを作成する必要がある場合、Oracle は優れた選択肢です。Oracleは様々なプラットフォームにわたってより包括的な検索を行うだけでなく、Webページや PowerPoint プレゼンテーションを生成する機能も活用します。これにより、必要な情報をより効果的に提示することができます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2F6qysYKWgJzr6O9cFQZy2%2FWebnew.gif&width=768&dpr=4&quality=100&sign=c1e5c777&sv=2) Oracle 生成の Web ページ さらに、コミュニティ内の専門家が提供するテンプレートを検索、閲覧、または活用することができます。これらのリソースを使用することで、より簡単かつ明確にあなたの疑問を解決することができます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FKmLmvdbROLspFLlX9Mf1%2FOracle-%25E4%25BB%2580%25E4%25B9%2588%25E6%25A0%25B7%25E7%259A%2584%25E9%2597%25AE%25E9%25A2%2598-3.%25E7%25A4%25BE%25E5%258C%25BA-edit.png&width=768&dpr=4&quality=100&sign=def7f672&sv=2) コミュニティで利用可能な様々な専門的テンプレート [PreviousOracle がサポートする現在の機能](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-gasaptosuruno) [NextOracle モード ユーザーガイド](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdo-yzgaido) Last updated 10 months ago Was this helpful? --- # flowith ナレッジベース・ユースケース | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) 以下は、クローズドベータテストとオープンベータテスト期間中に製品チームが収集した印象的な事例です： 1. **メディアコンテンツ制作（Content Creation）——コミュニティケーススタディ** Alexは、小紅書（RED）のコンテンツクリエイターで、質の高い投稿で多くのフォロワーを魅了してきました。しかし、完全なオリジナルコンテンツでは最新トレンドに追いつけず、成長が停滞していました。弊社のベータテスト期間中、彼は過去の執筆素材や記事を flowith knowledge base にアップロードしました。新しい記事を書く際、flowith が自動的に関連素材をマッチングし、彼の文体に基づいてパーソナライズされたコンテンツを迅速に生成できるようになりました。以前は毎回一から書き始める必要がありましたが、現在では入力部分で新しいテーマを設定するだけで、自分の文体に合わせた、個人の素材を活用した高品質な記事を素早く生成できるようになりました。彼の言葉によれば、作業時間が80%削減された」とのことで、現在は面白いアイデアの構想に専念できるようになっています。 1. **学生の試験対策・文書作成（Education & Academia）——パブリックベータユーザーの事例** Ze さんは期末試験の準備をしている際、ある講義の授業資料と参考文献を flowith ナレッジベースにインポートしました。システムは文書作成中にリアルタイムで必要な知識ポイントを提供します。Ze さんは、これらの機能を活用することで、論文の執筆効率が向上しただけでなく、試験勉強もスムーズで楽しくなりました。もう資料を苦労して探す必要がなく、学習がよりスマートで効率的になりました。 1. **ナレッジベースを活用したチームコラボレーション ―― パブリックベータユーザーの事例** ある教育テクノロジー企業のマーケティング部門では、年間を通じて異なるユーザー層向けに数百本のマーケティングコンテンツを制作する必要がありました。flowith の導入以前は、チームの素材が様々なプラットフォームに散在しており、検索や再利用の効率が低い状態でした。現在では、全てのマーケティング資料を flowith のナレッジベースに一元化することで、AI が関連素材を自動的に活用し、チームのスタイルに合わせたコンテンツを生成できるようになりました。これにより、重複作業が大幅に削減されただけでなく、より重要な点として、チームの制作物の品質とスタイルの統一性が向上しました。 1. **学術研究（Research）——パブリックベータユーザーの事例** あるベータテストユーザーである大学教授は、flowithを活用して学術キャリアで収集した資料を管理し、様々な情報源からのデータや文献を自動的に集約することで研究をサポートし、より広範な視野とより深い洞察を学術貢献にもたらしています。 1. **教師の授業準備——パブリックβテストユーザー事例** ある大学教師が次のような課題に直面していました：過去5年間に蓄積した百点以上の論文、教案、講義資料を統合し、新学期の一連の講義を準備する必要がありました。flowith に取り込んだ後、AI が自動的にこれらの資料を知識シード（Seeds）に分解し、関連付けを行いました。現在では講義テーマを入力するだけで、flowith が関連コンテンツを自動的にマッチングし、授業準備の効率が大幅に向上しています。授業準備の時間が60%削減され、むしろ講義の質が向上しました」と、この教師は述べています。 1. **ポーカー — コミュニティケーススタディ** コミュニティユーザーの Xavier はベータテストグループ内でポーカーに関するナレッジベースを公開しました。彼はプロプレイヤーの試合分析、戦略記事、実践例を収集し、ポーカーゲームにおける学習支援と高品質な意思決定を動的にサポートする「外部ツール」を開発しました。一般的なAIと比較して、はるかに優れた効果を発揮していることが確認できます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F2916067171-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252F2agAjJERs7Y6IiotYFD6%252Fuploads%252FijVTFeHBwo8IsoiR3Gio%252F%25E5%25BE%25B7%25E6%2589%2591%25E6%259B%25B4%25E6%2596%25B0.jpg%3Falt%3Dmedia%26token%3D1d3fff18-4b49-42aa-8872-0a3d1c43a3cd&width=768&dpr=4&quality=100&sign=2e62e64e&sv=2) 1. **大学出願―コミュニティケーススタディ** 教育コンサルタントが、Harvard、MIT、Yale などの一流大学への数百件の合格エッセイ例を体系化し、flowith Marketplace 上でナレッジベースとして公開しました。このナレッジベースは、これらの成功例の共通する特徴を分析できるだけでなく、志願者の背景に基づいてエッセイの構成をカスタマイズし、質の高い初稿を作成することができます。このナレッジベースは公開から1週間で800ドル以上の収益を上げました。 つまり、個人情報や知識の強化によってAIの生成効果を向上させることができる分野であれば、flowith ナレッジベースはあらゆる種類の知識労働に適用でき、現存の AI 製品の中で最高のクリエイティブ効果をもたらすことができます。 [Previousflowith ナレッジガーデンの概要](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdenno) [Nextナレッジベースの構築と使用方法](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/narejjibsunoto) Last updated 7 months ago Was this helpful? --- # Oracle がサポートする現在の機能 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FLb1TccN0scGPrEBlzdlD%2FAmazing%2520Mockups.jpeg&width=1248&dpr=4&quality=100&sign=50a313dc&sv=2) 以下は、flowith の Oracle モードが現在サポートしている外部ツールのリストです。このリストは継続的に更新されています。Oracle に特定のツールのサポートを希望される場合や、お客様の製品の API を Oracle と連携させることにご興味がある場合は、ぜひお問い合わせください。 **Tool Name** **Tool Function** analyze\_website\_traffic analyze specific website traffic; get website metrics; check website engagement ask\_user\_more ask user to input more information deep\_thinking deep analysis on a subject gen\_webpage generate web page, create website, design user interface, build online platform, develop web application, construct digital presence, craft internet portal, produce web-based solution, create ppt presentations, design infographics, generate charts and graphs, visualize data, create mindmaps and flowcharts. generate\_image generate images; draw images; text to image generations; draw a cute cat generate\_python help user run python code,code interpreter; do complex mathematical operations; do data visualization; run python code online\_search search realtime information by Google read\_website read a website's content; scrape website information; retrieve online articles search\_google\_scholar search academic papers; find scholarly articles; search\_images google image search; search\_jobs search for jobs search\_linkedin\_company search company on LinkedIn search\_linkedin\_people search people on LinkedIn, find LinkedIn profiles search\_patents search patents, find patent information, lookup inventions search\_products search for products, find items to buy, compare prices, shopping information search\_reddit\_posts search Reddit posts search\_stackexchange\_question search stackoverflow posts search\_twitter search tweets, find twitter posts search\_videos search YouTube videos, video search search\_websites search for websites, find links, google search search\_xiaohongshu\_notes search xiaohongshu posts send\_email send email, send mail, g-mail, contact a person by e-mail send\_rapid\_email send email; fast email delivery summarize\_and\_decide summarize contents translate\_text translate content user provided [PreviousOracle モードについて](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdonitsuite) [NextOracle モードはどのような問題解決に適していますか？](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdohadonoyounanishiteimasuka) Last updated 10 months ago Was this helpful? --- # ご契約に関して | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) [](https://doc.flowith.io/flowith-heyoukoso/sononogo/gonishite#sabusukuripushonpuran) **サブスクリプションプラン** ----------------------------------------------------------------------------------------------------------- 弊社では、費用対効果の高いサブスクリプションプランをご用意し、OpenAI、Anthropic、Google、DeepSeek、Meta などの企業が提供する最先端の AI モデルをサポートしております。さらに、画像や動画を生成できる最先端のモデルも導入しております。 詳細については、以下のページをご覧ください：[https://flowith.io/sub](https://flowith.io/sub) [](https://doc.flowith.io/flowith-heyoukoso/sononogo/gonishite#spmoderu-oracle) **スーパーモデル — Oracle** --------------------------------------------------------------------------------------------------------- Oracle モードでは、各ステップごとに1つの T-1 クォータを消費します。例えば、6 つのステップを必要とするタスクの場合、6 つの T-1 クレジットを使用することになります。 [](https://doc.flowith.io/flowith-heyoukoso/sononogo/gonishite#narejjibsu) **ナレッジベース** ------------------------------------------------------------------------------------------- 1. ファイル保存料金体系 ナレッジベースにファイルをアップロードする際、ストレージ容量の割り当てのみが計算され、追加のモデル使用料は発生いたしません。 1. 対話サービス料金規定 ナレッジベースを通じて対話を開始する場合、選択したモデルのサービスレベル（T1レベル、T2レベル）に応じてモデル使用回数が計測され、基本対話サービスと同様の料金体系で課金されます。 1. ストレージ割り当て管理 ナレッジベースのストレージ割り当ての具体的な制限事項および管理規定については、以下をご参照ください： [https://flowith.io/sub](https://flowith.io/sub) [](https://doc.flowith.io/flowith-heyoukoso/sononogo/gonishite#go) **ご注意事項：** ---------------------------------------------------------------------------------- **ポリシーについて：** 一部のユーザーが大量の Flowith アカウントを一括登録し、高コストの AI モデルの無料利用枠を悪用する事例が確認されております。この行為は他のユーザーの通常利用に重大な影響を及ぼしています。 利用制限に達した場合で、特別な事情がございましたら、当社までご連絡ください。 連絡先： email：[\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection) Discord：[https://discord.com/invite/g95B77xbpD](https://discord.com/invite/g95B77xbpD) ### [](https://doc.flowith.io/flowith-heyoukoso/sononogo/gonishite#porish) 返金ポリシー * サブスクリプション料金は原則として返金不可となります * 以下の場合のみ返金申請が可能です： * 支払い完了後48時間以内にご連絡ください。 * サブスクリプション開始後の累計使用クレジットが200未満の場合。 以下の場合は、[\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection#c5b6b0b5b5aab7b185a3a9aab2acb1adeba6aa) までご連絡ください： • その他のご不明点や特別な状況について • 返金条件を満たす申請について • サブスクリプション解約に関する技術的なお問い合わせ [PreviousOracle モードの高度なテクニック](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdononatekunikku) [Next問題の報告方法](https://doc.flowith.io/flowith-heyoukoso/sononogo/no) Last updated 2 months ago Was this helpful? --- # 高度な機能 — ノード参照 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/na-ndo#ndotoha) **ノードとは？** ---------------------------------------------------------------------------------------- ノードは、Flowith キャンバス上の最小単位です。AI の応答とユーザーの入力それぞれが1つのノードを構成します。 [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/na-ndo#ndonoi) **変数ノードの使い方** ------------------------------------------------------------------------------------------ 変数ノード参照は、頻繁に変更される内容（翻訳が必要なテキストなど）に最適です。 ### [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/na-ndo#purntekisuto) プレーンテキスト ノードの挿入 1. 左上のツールバーにある 「コメントを追加」 ボタンをクリックし、キャンバス上にドラッグします。 2. ノードの名前を変更します。 3. 内容を入力します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FtDJ2opVltrN32eiRhTyR%2F%25E6%25B5%2581%25E5%25BC%258F-%25E9%25AB%2598%25E7%25BA%25A7%25E7%258E%25A9%25E6%25B3%2595-%25E7%25BA%25AF%25E6%2596%2587%25E6%259C%25AC%25E6%258F%2592%25E5%2585%25A5%25E8%258A%2582%25E7%2582%25B9.gif&width=768&dpr=4&quality=100&sign=7915480f&sv=2) キャンバス上のノードの挿入 他のノードを参照する方法 1. 挿入したい位置にカーソルを移動します 2. キャンバス下部にある円形のボタンをクリックします。 3. 表示されるメニューから 「他のノードの内容を参照」 を選択します。 4. 参照したいノードをクリックして選択します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2F387H3Fqw17eG4tY6HAYs%2F%25E6%25B5%2581%25E5%25BC%258F-%25E9%25AB%2598%25E7%25BA%25A7%25E7%258E%25A9%25E6%25B3%2595-%25E5%25BC%2595%25E7%2594%25A8%25E8%258A%2582%25E7%2582%25B9.gif&width=768&dpr=4&quality=100&sign=6bd625da&sv=2) ノードの参照を選択できるときにキャンバスが暗くなります ### [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/na-ndo#pdf-yanoappurdo) PDF や画像のアップロード **PDF をアップロードする** 1. カーソルを挿入したい位置に配置します。 2. キャンバスの下にある円形のボタンをクリックします。 3. 「PDFをアップロード」 を選択します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FdITfobhKgasDyIr9MnZc%2F%25E6%25B5%2581%25E5%25BC%258F-%25E9%25AB%2598%25E7%25BA%25A7%25E7%258E%25A9%25E6%25B3%2595-%25E4%25B8%258A%25E4%25BC%25A0%25E6%2596%2587%25E6%259C%25AC%25E6%2596%2587%25E4%25BB%25B6.gif&width=768&dpr=4&quality=100&sign=c3253f8b&sv=2) テキストファイルは、キャンバス上に 「挿入されたノード」 として表示されます **画像ファイルのアップロード方法** 1. 挿入したい位置にカーソルを合わせます。 2. キャンバス下部にある円形のボタンをクリックします。 3. 「画像をアップロード」 を選択します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FoMcwTphQ5BV0suVRwjDb%2F%25E6%25B5%2581%25E5%25BC%258F-%25E9%25AB%2598%25E7%25BA%25A7%25E7%258E%25A9%25E6%25B3%2595-%25E4%25B8%258A%25E4%25BC%25A0%25E5%259B%25BE%25E7%2589%2587.gif&width=768&dpr=4&quality=100&sign=d1e49383&sv=2) アップロードされた画像は 「挿入ノード」 としてキャンバスに表示されます [PreviousFlow モード ユーザーガイド](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdo-yzgaido) [Nextプロンプトエンジニアリングの基礎](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/puronputoenjiniaringuno) Last updated 10 months ago Was this helpful? --- # 知识花园变现指南 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) AI 时代的知识付费生态将迎来全新的范式转变，传统基于图文和视频的知识创作将慢慢被淘汰，基于 AI 的知识创作和分发将成为未来的主流。flowith 作为当前海内外市场上用户量最大的画布式 AI 生成工具于 2025 年推出了 「知识市场 Knowledge Marketbase」 ——我们希望 flowith 能成为 AI 时代知识创造者的 「OnlyFans」 ，帮助创作者收获更多的收益和曝光，给使用者提供更多价值。 目前，已经有我们合作的知识博主在内测期间**通过发布其原创的财富管理知识库，在短短两周内获得了超过 2000 位用户的测试，并创造了超过 2 万元的收入。** [](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-bian-xian-zhi-nan#shen-me-shi-flowith-zhi-shi-shi-chang) 什么是 flowith 知识市场 --------------------------------------------------------------------------------------------------------------------------------------------- ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FIKBz2MIWVN8wPnQdsQ4O%252F%25E7%259F%25A5%25E8%25AF%2586%25E5%25B8%2582%25E5%259C%25BA.jpg%3Falt%3Dmedia%26token%3Dc79b71c9-5ed9-4215-979a-ea1b0ab2ae1c&width=768&dpr=4&quality=100&sign=9845652&sv=2) 尽管大语言模型在大规模通用语料上已学到大量知识，但其参数记忆有限且难以及时获取最新、最具体、私有化的专业信息。通过在使用中引入外部知识库，可动态补充和检索所需内容，减少模型幻觉，提升准确度及可控性，显著增强 LLM 的生成质量与应用价值。联通 AI 时代的知识管理和创作，flowith 通过自研的 agenic knowledge management 框架可以将用户上传的信息和文件高效拆分为 knowledge seed，并使用自动化的知识管理进行结构化的存储和整理。在用户后续使用 AI 时可以自动匹配相关的知识种子，从而大幅度提高 AI 的回复质量、针对性、有效性。 简而言之，你可以将 flowith 知识库理解为大语言模型的 「知识外挂」 ，通过引入这些 「外挂」 ，大模型可以拥有你和团队的独家信息，获得当前市面上所有 AI 生成类产品中所能得到最高质量的生成和创作效果。而知识市场就是这些知识库的市集，允许用户分享免费的知识库或使用知识库进行内容变现。 [](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-bian-xian-zhi-nan#ru-he-chuang-jian-yi-ge-hao-de-zhi-shi-ku) 如何创建一个好的知识库 -------------------------------------------------------------------------------------------------------------------------------------------- 像几乎任何领域的内容创作一样，要在 flowith 上创建一个可以给你的用户提供价值并为你自己带来收益的知识库有两种方法： 1. **原创（Creation）**：如果你已经是一位知识和内容创作者或者博主，太好了！除了传统基于图片、文字、视频的内容分发，flowith 为你提供了一条额外通过内容变现的方式：AI。你只需要将自己的原创内容按照不同方式导入 flowith 就可以进行分享和上架。 2. **策展（Curation）：**在这个信息繁杂丰富的时代，互联网已经为每个人提供了足够多的素材和优质内容，难点并不在于缺乏信息，而在于筛选和整理其中有洞见和价值的那些。你可以将互联网上的优质内容按照特定主题进行整理和收集并上传到 flowith 知识库当中，就可以开始使用内容变现了。 [](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-bian-xian-zhi-nan#yuan-wen-jian-shang-chuan-zhi-nan) 源文件上传指南 -------------------------------------------------------------------------------------------------------------------------------- ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252Fory1suXUnPBj75mjprxi%252Fzhong%25201.jpg%3Falt%3Dmedia%26token%3D049e660f-fee5-42d4-9991-ce80c2d1505b&width=768&dpr=4&quality=100&sign=8156128b&sv=2) * **长度要求：** * 内容长度灵活，可处理包含相关关键词的任意文本 * 不支持文本摘要生成功能 * 超过1,000字符的源文本将自动分段处理 * **文件格式规范：** * 无特定格式限制，但不支持PDF图像识别 * 图片需单独上传，嵌入长文本中的图片将被忽略 * 不支持代码的读取和处理 * **支持的内容类型：** * 支持单独的图片上传 * 完全支持基于文本的内容 * 长篇内容需要分段处理 [](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-bian-xian-zhi-nan#zai-zhi-shi-shi-chang-fa-bu-he-ying-xiao-ni-de-zhi-shi-ku) 在知识市场发布和营销你的知识库 ---------------------------------------------------------------------------------------------------------------------------------------------------------------- ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FHdxB88TKt77RojJ57AeQ%252F%25E7%259F%25A5%25E8%25AF%2586%25E5%25B8%2582%25E5%259C%25BA2.jpg%3Falt%3Dmedia%26token%3Dcaaff5d8-6621-47fa-a8fb-70add644dbe8&width=768&dpr=4&quality=100&sign=ea22b61a&sv=2) 创建一个好的知识库只是第一步：Knowledge Market 是用户进行浏览和购买知识库的平台，flowith 的大量的社区用户在此寻觅适合自己、能够解决自己问题的知识库，并为此付费。明晰且具有吸引力的名字和描述可以帮助你的知识库脱颖而出，获得更多用户的青睐——自然帮助你获取更多潜在收益。 ### [](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-bian-xian-zhi-nan#zhi-shi-ku-miao-shu-zen-me-xie) 知识库描述怎么写 #### [](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-bian-xian-zhi-nan#yi-qi-yi-ge-hao-ming-zi) 一、起一个好名字 * 单一信源知识库 如果是单一信源的知识库，我们建议直接使用信息源的名字：如孙子兵法 * 合集、专栏、复合型知识库 如果你的知识库是一个复合型的知识库，可以选择一个能清晰简洁地描述整体主题的标题。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FXeq5kmz3zMBxMHHlVaHJ%252Fzhong%25202.jpg%3Falt%3Dmedia%26token%3D3c0fa4bf-1437-41e0-b10b-3f27cd8dc32b&width=768&dpr=4&quality=100&sign=8b6543ea&sv=2) #### [](https://doc.flowith.io/cn/knowledge-garden/zhi-shi-hua-yuan-bian-xian-zhi-nan#er-zhuan-xie-hao-de-miao-shu) 二、撰写好的描述 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FKSLivUkeEVFPmfIRdWnq%252Fzhong%25203.jpg%3Falt%3Dmedia%26token%3D66a9bc51-9dc6-4f0a-ad4e-d3945b412afb&width=768&dpr=4&quality=100&sign=bcb19f8&sv=2) 好的知识库描述是你的门面，能让用户快速理解其价值和用途。根据经验，我们建议你从以下四个方面进行说明： 1. **信息来源：**明确说明知识内容的来源渠道、标注信息的时效性和更新频率、注明作者或维护团队的背景 2. **知识库特点：**概述知识库的主要内容范围、说明知识组织的方式和结构、突出知识库的独特优势或特色 3. **目标用户：**明确定义适用的用户群体、描述用户可能的使用场景 4. **价值主张：**指出这个知识库能解决什么具体问题、说明用户可以获得什么收益、描述预期达到的学习或应用效果 Copy 例如: 这是一个面向前端开发者的 React 最佳实践知识库。内容来自我们团队3年的项目实践总结，每月更新。 通过系统化的组件开发指南和性能优化建议，帮助开发者提升代码质量，加快项目开发效率。 记住：描述要简洁清晰，让用户在短时间内就能判断这个知识库是否适合自己。同时要确保描述的真实性，不要过度承诺。 [Previous如何建立和使用知识库](https://doc.flowith.io/cn/knowledge-garden/ru-he-jian-li-he-shi-yong-zhi-shi-ku) [Next关于 Oracle 模式](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/guan-yu-oracle-mo-shi) Last updated 5 months ago Was this helpful? --- # Flow モードの紹介 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2F3GtBbyWKKz2HSlW0MPft%2F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE2.png&width=1248&dpr=4&quality=100&sign=99caed33&sv=2) flow モードは、コンテンツクリエイター向けに特別に設計された次世代の AI 対話方式です。これは、flowith における標準的な対話方法として機能します。flowith は無限に広いキャンバスを活用しており、複数の AI アシスタントと同時にやり取りしながら、必要なコンテンツを生成することができます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2F3GtBbyWKKz2HSlW0MPft%2F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE2.png&width=768&dpr=4&quality=100&sign=99caed33&sv=2) **なぜ flow モードを使用するのか？** 単純な翻訳や基本的な質問応答だけで AI を利用する場合、従来のチャットモードで十分かもしれません。しかし、AI との多段階の会話や、以前の入力の修正、あるいは複数の AI アシスタントとの同時協働が必要な場合、flow モードは全く異なる創造的な体験を提供します。 flow モードを使用すると、より複雑なタスクや長期的なプロジェクトに対して、AI とのインタラクションを柔軟に管理できます。これにより、より効率的で生産的な作業環境を実現し、創造性を最大限に引き出すことができます。 **flow モードの利点：** 1. **複数の AI インスタンスが並列スレッドで同時にコンテンツを生成できます。** 2. **AI プロンプトを繰り返し修正し、複数回生成することができます。** 3. **各ノードが他のノードのコンテンツを「参照」できるため、コンテンツの再利用が可能です。** 4. **柔軟な情報の分岐と統合ができます。** 5. **不要な情報をその場で削除できます。** 6. **異なるモデルの生成結果を並行して簡単に比較できます。** 7. ... #### [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdono#ysuksu) ユースケース: > flow モードでは、新しい会話を始めることなく、キャンバス内でモデルを簡単に切り替えることができます。例えば、「9.11と9.9のどちらが大きいか？」 という質問に対して、同じキャンバス上で複数の主要モデルの回答を比較できます。flowith の一般モードでは、ワンクリックでモデルを切り替えるだけでなく、各モデルのこの質問に対するパフォーマンスをキャンバス内に明確に表示します。これにより、結果をより効果的に比較することが可能になります。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FsAWfdIx0IQSPTVHDsT7N%2F%25E6%25B5%2581%25E5%25BC%258F%25E4%25BB%258B%25E7%25BB%258D-%25E6%25A1%2588%25E4%25BE%258B-%25E4%25B8%2580%25E9%2594%25AE%25E5%2588%2587%25E6%258D%25A2%25E6%25A8%25A1%25E5%259E%258B.gif&width=768&dpr=4&quality=100&sign=28265d65&sv=2) ワンクリックでモデル切り替え > 1つのキャンバス内で、モデルの切り替えだけでなく、モード間を自由に切り替えることができます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Frj5DflZVr88MBoqvu4FM%2F%25E6%25B5%2581%25E5%25BC%258F%25E4%25BB%258B%25E7%25BB%258D-%25E6%2596%2587%25E6%259C%25ACmode%25E5%2588%2587%25E6%258D%25A2%25E5%259B%25BE%25E7%2589%2587mode.gif&width=768&dpr=4&quality=100&sign=d98131d4&sv=2) 「テキスト生成」 から 「画像生成」 へスムーズに移行 さらに、一般モードではテキストファイルや画像ファイルのアップロードにも対応しています。アップロードしたテキストファイルや画像ファイルに関連する質問をすることができます。 [高度な機能 — ノード参照](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/na-ndo) * * * Flowの使用方法や機能に関する**追加情報**を、**継続的に更新**してまいります。 [PreviousFlowith へようこそ](https://doc.flowith.io/flowith-heyoukoso) [NextFlow モード ユーザーガイド](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdo-yzgaido) Last updated 10 months ago Was this helpful? --- # 情報収集＋整理のマエストロ | Flowith Official Document 数十、数百の AI Agents を同時に活用できるようになりました。 Oracle は、AutoGPT よりも実用的で強力なエージェントフレームワークです。自律的にタスクを計画し、ワークフローを設計し、ツールを呼び出し、後続のワークフローを反復的に改善します。 [Previous問題の報告方法](https://doc.flowith.io/flowith-heyoukoso/sononogo/no) Last updated 10 months ago Was this helpful? --- # 問題の報告方法 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) 製品使用中に問題が発生した場合や改善のご提案がある場合は、ぜひフィードバックをお寄せください。皆様からのご意見を心よりお待ちしております。 **お問い合わせ方法：** 公式メールアドレス： [\[email protected\]](https://doc.flowith.io/cdn-cgi/l/email-protection) **製品ページからのお問い合わせ:** 1. flowith.ioの製品ページ左下にある 「サプスクリプションと設定」 ボタンをクリックします。 2. 「サプスクリプションと設定」 メニューから 「フィードバック」 を選択します。 3. ご希望のフィードバックの種類を選んでください。 [Previousご契約に関して](https://doc.flowith.io/flowith-heyoukoso/sononogo/gonishite) [Next情報収集＋整理のマエストロ](https://doc.flowith.io/flowith-heyoukoso/ysuksu/nomaesutoro) Last updated 10 months ago Was this helpful? --- # Oracle 模式使用指南 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FIoGBv6KmDuBGVIrYtsfM%252F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE4.jpg%3Falt%3Dmedia%26token%3Dc51d1b64-dc29-4d04-a197-e95cb7aa7124&width=1248&dpr=4&quality=100&sign=bbf8cc9c&sv=2) [](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mo-shi-shi-yong-zhi-nan#kai-shi) 开始 ----------------------------------------------------------------------------------------------------------- 1. 在对话框中输入问题 2. 打开 Oracle 模式 3. 点击 Start 4. 查看 planner 并点击 「Confirm & Run」 执行 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FjlrwtzJoOQ94RMVOrbNx%252FOracle%25E5%25BC%2580%25E5%25A7%258B%25E5%2588%259B%25E5%25BB%25BA.gif%3Falt%3Dmedia%26token%3D4ca1ed03-f72f-4fad-9af8-8c6f8d0c01fb&width=768&dpr=4&quality=100&sign=b48726af&sv=2) 「ORACLE MODE」 要保持打开 [](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mo-shi-shi-yong-zhi-nan#cha-kan-recipe-de-zhi-xing-jin-du) 查看 Recipe 的执行进度 -------------------------------------------------------------------------------------------------------------------------------------------------- 1. 选中一个节点 2. 将光标滑动到页面下方输入框 3. 查看 Recipe 的执行进度 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252F7YffjwPs8fosNtW18bRB%252F%25E6%259F%25A5%25E7%259C%258BPlanner.gif%3Falt%3Dmedia%26token%3D87e398d3-6840-49df-ac00-e8c7230176b0&width=768&dpr=4&quality=100&sign=81db7773&sv=2) 每一条步骤前的圆圈表明该条步骤的状态 [](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mo-shi-shi-yong-zhi-nan#chong-fu-zhi-xing-shang-yi-ge-recipe) 重复执行上一个 Recipe ---------------------------------------------------------------------------------------------------------------------------------------------------- 1. 点击你想查看的 recipe 的任意一个节点 2. 将光标移动到对话框上，recipe 会自动弹出 3. 点击 「Re-run」，会弹出确认页面 4. 点击 「Confirm & Run」 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FwNqNDtPaqKKQYZReRhbd%252F%25E9%2587%258D%25E5%25A4%258D%25E6%2589%25A7%25E8%25A1%258C.gif%3Falt%3Dmedia%26token%3D21cb2613-04e5-44f4-adf5-e0e952c7e4ec&width=768&dpr=4&quality=100&sign=69a04a9b&sv=2) 重复执行上一个 recipe [](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mo-shi-shi-yong-zhi-nan#ru-he-zai-tong-yi-ge-hua-bu-zhong-kai-qi-xin-fen-zhi) 如何在同一个画布中开启新分支 -------------------------------------------------------------------------------------------------------------------------------------------------------------------- 点击 「New Branch」 后，在页面下方的对话框中输入新问题即可开启新分支。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FnDA7SRBSXcnh0vf6AJrM%252F%25E5%25BC%2580%25E5%2590%25AF%25E6%2596%25B0%25E5%2588%2586%25E6%2594%25AF.gif%3Falt%3Dmedia%26token%3D3ef4f015-7f8d-4c63-b532-ef4f59fd5dab&width=768&dpr=4&quality=100&sign=c261fec3&sv=2) 一个画布中可以开启多个分支 [](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mo-shi-shi-yong-zhi-nan#ru-he-shi-yong-recipe-mu-ban) 如何使用 Recipe 模板 -------------------------------------------------------------------------------------------------------------------------------------------- 1. 在 「发现」 搜索想要的模板关键字 2. 点击 「Use this recipe」 查看详情 3. 点击 「Confirm & Run」 即可使用模板 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FaB5P8Xlsn6N0QuSHrv7v%252F%25E4%25BD%25BF%25E7%2594%25A8%25E6%25A8%25A1%25E6%259D%25BF.gif%3Falt%3Dmedia%26token%3Dd38b91a9-47ec-4f1f-87d0-4c22a63739f3&width=768&dpr=4&quality=100&sign=a768af2e&sv=2) 社区中有专业的模板可供选择 [PreviousOracle 适合解决什么样问题](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-shi-he-jie-jue-shen-me-yang-wen-ti) [NextOracle 模式高级玩法](https://doc.flowith.io/cn/oracle-mode-zhi-neng-ti-mo-shi/oracle-mo-shi-gao-ji-wan-fa) Last updated 11 months ago Was this helpful? --- # Oracle モードについて | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fzu8C1nUY5LDCaMxYYNnm%2F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png&width=1248&dpr=4&quality=100&sign=4dfcd71&sv=2) [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdonitsuite#oracle-toha) **Oracle とは？** -------------------------------------------------------------------------------------------------------------------- Oracle は、flowithチームが開発した最先端のインテリジェントエージェントシステムです。pre-AGI（pre-Artificial General Intelligence）の実現を目指して設計されました。複数のツールを使用する必要がある多段階の複雑なタスクを処理するために特別に設計されています。 Oracleの主な特長は以下の通りです： * **自律的なタスク計画** * **タスクの分解** * **無制限のツール活用** * **自己最適化** Oracleの外部機能は継続的に更新・強化されており、時間とともにより強力になっていきます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FLb1TccN0scGPrEBlzdlD%2FAmazing%2520Mockups.jpeg&width=768&dpr=4&quality=100&sign=50a313dc&sv=2) Oracle Recipe and Nodes つまり、Oracleはデータレベルで動作する汎用エージェントです。これは以下の特徴を持ちます： 1. データ層で機能し、物理的な媒体や実世界との直接的なやりとりは行いません。 2. 汎用性が高く、事前に訓練された機能や定義された機能に限定されず、幅広いタスクを実行できます。 3. 単なる大規模言語モデル（LLM）を超えたエージェントとして、ツールの利用、メモリの管理、複数の LLM との協調が可能です。 4. 単純なクエリやコマンドを超えた複雑なタスクを完了できます。 5. 新しい状況に適応する能力があり、未知の環境やタスクにおいても自己学習と処理が可能です。 [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdonitsuite#zhu-yao-gai-nian) 主要概念 --------------------------------------------------------------------------------------------------------------- **Recipe Workflow** 「recipe」 という言葉は、本来、料理の作り方を示す一連の手順を指します。Oracle においては、recipe は AI がユーザーのタスクを完了するために実行する体系的なワークフローを表します。ユーザーが要求や指示を入力すると、Oracle はその全体的な指示を、自身の能力と利用可能なツールに基づいて、より小さなサブタスクに分解します。これらのサブタスクは、その後、タスクアロケーターに渡され、段階的に実行されます。 完全な recipe を設計した後、Oracle は必ず実行を開始する前にユーザーに確認を取ります。この確認プロセス中、ユーザーはレシピ内の特定のステップを直接修正することができます。以下の操作が可能です： * Adjusting the order of steps * Deleting steps * Modifying text * Adding new steps ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fzu8C1nUY5LDCaMxYYNnm%2F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png&width=768&dpr=4&quality=100&sign=4dfcd71&sv=2) A Recipe Generated by Oracle **Dispatcher** Dispatcher 別名：タスク割り当て器）は、オラクルが生成したサブタスクの実行において重要な役割を果たす要素です。オラクルが生成した各サブタスクに対し、Dispatcher 内の専用エージェントが割り当てと実行を担当します。このエージェントは、オラクルがサポートするツールセットの中から最適なツールを選択し、サブタスクを実行して完了させます。 例えば、サブタスクが 「ソーシャルメディアで flowith のユーザーレビューを検索する」 場合、タスク割り当て機能は 「Twitter検索ツール」、「Reddit検索ツール」、そして 「Google検索ツール」 を呼び出し、Flowith のユーザーレビューを検索します。 Oracle の会話開始ページで、現在サポートされているツールの一覧をご確認いただけます。このリストはリアルタイムで更新されており、私たちは継続的に Oracle のサポートされているツールキットに最も有用なツールを追加しています。 [Oracle がサポートする現在の機能](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-gasaptosuruno) ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fg6n2Kda04BzP3TPTHtPJ%2F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE4.jpg&width=768&dpr=4&quality=100&sign=ebd383ce&sv=2) Tools supported by Oracle **Memory and Context Management** Oracle is designed for complex tasks, which often generate extensive context due to their multi-step nature. If we were to provide all previous context to the LLM (Large Language Model), it could potentially impact its performance. To address this, Oracle employs intelligent context management during task allocation and execution. For example, when Oracle reaches step 10 of a process, its memory manager evaluates which information from previous steps is essential and incorporates only the necessary context into the current step. This approach eliminates concerns about excessively long context. **Self-Iteration and Update Recipe** After completing each step, Oracle automatically updates the entire Recipe based on the execution results of each step. For instance, if it fails to find the required information, it will incorporate additional search steps into the subsequent stages to ensure successful task completion. When Oracle inquires about user satisfaction with the results, and the user expresses dissatisfaction, it will generate additional steps in an attempt to provide the desired outcome. [Previousflowith ナレッジガーデン収益化ガイド](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdengaido) [NextOracle がサポートする現在の機能](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-gasaptosuruno) Last updated 10 months ago Was this helpful? --- # flowith ナレッジガーデン収益化ガイド | Flowith Official Document AI 時代のナレッジエコノミーは新たなパラダイムシフトを迎えようとしています。従来の文章や動画によるコンテンツ制作は徐々に淘汰され、AI 駆動型のナレッジ創造と配信が主流となっていくでしょう。キャンバス型AI生成ツールとして国内外で最大のユーザー数を誇る flowith は、2025年に 「 Knowledge Marketbase（ナレッジマーケットベース）」を立ち上げます。私たちは、flowith がAI時代のクリエイターにとって 「 OnlyFans 」 のような存在となり、クリエイターの収益と露出を最大化し、ユーザーにより大きな価値を提供することを目指しています。 現在、ベータテスト期間中に、当社と提携しているナレッジクリエイターが、独自の資産運用に関するナレッジベースを公開することで、わずか2週間で2,000人以上のユーザーテストを獲得し、2万元以上の収益を上げています。 [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdengaido#flowithnarejjimkettotoha) Flowithナレッジマーケットとは --------------------------------------------------------------------------------------------------------------------------------------------------- ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F2916067171-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252F2agAjJERs7Y6IiotYFD6%252Fuploads%252FrPZ0H4DKeTj1nVGKq2LX%252FEn%2520MARKET.jpg%3Falt%3Dmedia%26token%3D00c596ee-06ee-42a3-9efa-2df934e330d7&width=768&dpr=4&quality=100&sign=994e5c0b&sv=2) 大規模言語モデルは汎用的なコーパスから膨大な知識を学習しているものの、パラメータによる記憶には限界があり、最新かつ具体的な専門情報やプライベートな情報をリアルタイムに取得することは困難です。外部知識ベースを利用することで、必要なコンテンツを動的に補完・検索し、モデルの幻覚を軽減し、精度と制御性を向上させ、LLMの生成品質と実用価値を大幅に向上させることができます。 AI 時代のナレッジマネジメントと創造をつなぐ flowith は、独自開発した agenic knowledge management フレームワークにより、ユーザーがアップロードした情報やファイルを効率的にknowledge seed に分解し、自動化されたナレッジマネジメントによって構造化された保存・整理を行います。ユーザーが後続のAI利用時に関連する知識シードを自動的にマッチングすることで、AI の応答品質、的確性、有効性を大幅に向上させることができます。 簡単に言えば、flowith ナレッジベースは大規模言語モデルの「外部知識拡張」として理解することができます。この 「 拡張機能 」 を導入することで、大規模モデルはあなたやチームの独自情報を活用でき、現在市場に存在するAI生成製品の中で最高品質のコンテンツ生成および創作効果を実現することができます。そして、ナレッジマーケットはこれらのナレッジベースが集まる市場として、ユーザーが無料でナレッジベースを共有したり、ナレッジベースを活用してコンテンツの収益化を行うことができます。 [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdengaido#retanarejjibsunori) 優れたナレッジベースの作り方 ----------------------------------------------------------------------------------------------------------------------------------------- あらゆる分野のコンテンツ制作と同様に、flowithで価値のある知識ベースを作成し、ユーザーに価値を提供しながら収益を上げる方法には、主に2つのアプローチがあります： 1. **オリジナルコンテンツの制作（Creation）**：すでにナレッジクリエイターやコンテンツクリエイター、ブロガーとしての活動をされている方には、これは最適な選択です！従来の画像、テキスト、動画によるコンテンツ配信に加えて、flowithではAIを活用した新しい収益化の方法を提供しています。自身のオリジナルコンテンツをflowithに様々な形式でインポートするだけで、共有や出品が可能になります。 2. **コンテンツキュレーション（Curation）**：この情報過多の時代において、インターネットは既に十分な素材と質の高いコンテンツを提供しています。課題は情報不足ではなく、洞察力があり価値のあるコンテンツを選別し整理することにあります。インターネット上の質の高いコンテンツを特定のテーマに沿って整理・収集し、flowithの知識ベースにアップロードすることで、コンテンツの収益化を開始することができます。 [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdengaido#ssufairunoappurdo) ソースファイルのアップロード手順 ------------------------------------------------------------------------------------------------------------------------------------------ ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F2916067171-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252F2agAjJERs7Y6IiotYFD6%252Fuploads%252FUB9i4ZwpyyiaNOMRgcY7%252F2.jpg%3Falt%3Dmedia%26token%3D24c14c9f-45df-4050-8a0e-d6580f852a8a&width=768&dpr=4&quality=100&sign=698e39a9&sv=2) * 長さの要件： * コンテンツの長さは柔軟で、関連キーワードを含む任意のテキストを処理可能 * テキスト要約機能には対応していません * 1,000文字を超えるソーステキストは自動的に分割して処理されます * ファイル形式の仕様： * 特定の形式制限はありませんが、PDFの画像認識には対応していません * 画像は個別にアップロードする必要があり、長文テキスト内に埋め込まれた画像は無視されます * コードの読み取りおよび処理には対応していません * 対応コンテンツタイプ： * 単独の画像アップロードに対応 * テキストベースのコンテンツに完全対応 * 長文コンテンツは分割処理が必要 [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdengaido#narejjibsunotomketingu) ナレッジベースの公開とマーケティング ------------------------------------------------------------------------------------------------------------------------------------------------- ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F2916067171-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252F2agAjJERs7Y6IiotYFD6%252Fuploads%252Fa3q2YlKbdsU2a3tVd7eE%252F%25E5%25B8%2582%25E5%259C%25BAEn2.jpg%3Falt%3Dmedia%26token%3Db659dab3-a3b5-4fac-b613-6b6a76cf1c57&width=768&dpr=4&quality=100&sign=d29462cf&sv=2) 優れたナレッジベースの作成は第一歩に過ぎません：Knowledge Market はユーザーがナレッジベースを閲覧・購入できるプラットフォームです。flowith の多くのコミュニティユーザーは、自分に適した問題解決が可能なナレッジベースを探し、対価を支払っています。明確で魅力的な名称と説明文があれば、あなたのナレッジベースは際立ち、より多くのユーザーから支持を得ることができます——それは自然とより多くの収益につながります。 ### [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdengaido#narejjibsunonoki) ナレッジベースの説明文の書き方 #### [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdengaido#id-1-iwokeru) 1\. 良い名前を付ける * 単一情報源のナレッジベース 単一情報源のナレッジベースの場合、情報源の名称をそのまま使用することを推奨します（例：孫子の兵法） * コレクション・専門コラム・複合型ナレッジベース ナレッジベースが複合的な性質を持つ場合は、全体のテーマを明確かつ簡潔に表現できるタイトルを選択してください。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F2916067171-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252F2agAjJERs7Y6IiotYFD6%252Fuploads%252FT0U9XK3XV8X93blgi27F%252F1.jpg%3Falt%3Dmedia%26token%3Dee93da9c-ab17-450c-ae32-0a3e683a73ad&width=768&dpr=4&quality=100&sign=138826be&sv=2) #### [](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdengaido#id-2-retano) 2\. 優れた説明文の作成 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F2916067171-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252F2agAjJERs7Y6IiotYFD6%252Fuploads%252FGXTOS0qyfjraJoLC7iHn%252F3.jpg%3Falt%3Dmedia%26token%3D7937f760-e4de-47ff-a59c-9b0f6f91cafe&width=768&dpr=4&quality=100&sign=58d98f36&sv=2) 優れたナレッジベースの説明は**あなたの顔**となり、ユーザーがその価値と用途を素早く理解できるようになります。経験則として、以下の**4つの観点**から説明することをお勧めします： * **情報源：**知識コンテンツの出所を明確に示し、情報の有効期限と更新頻度を明記し、著者またはメンテナンスチームの背景を記載 * **ナレッジベースの特徴：**ナレッジベースの主な内容範囲を概説し、知識体系の構成方法と構造を説明し、ナレッジベースの独自の強みや特色を強調 * **対象ユーザー：**適用対象となるユーザー層を明確に定義し、想定される利用シーンを説明 * **価値提案：**このナレッジベースが解決できる具体的な課題を指摘し、ユーザーが得られる利点を説明し、期待される学習・活用効果を記述 Copy 例えば： これはフロントエンド開発者向けの React ベストプラクティス知識ベースです。私たちチームの3年間のプロジェクト実践から得られた内容を毎月更新しています。 体系的なコンポーネント開発ガイドラインとパフォーマンス最適化の提案を通じて、開発者のコード品質向上とプロジェクト開発効率の改善をサポートします。 **覚えておいてください**：説明は簡潔かつ明確にし、ユーザーが短時間でこのナレッジベースが自分に適しているかどうかを判断できるようにしてください。また、説明内容の真実性を確保し、過度な約束は避けてください。 [Previousナレッジベースの構築と使用方法](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/narejjibsunoto) [NextOracle モードについて](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdonitsuite) Last updated 7 months ago Was this helpful? --- # Oracle モード ユーザーガイド | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fg6n2Kda04BzP3TPTHtPJ%2F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE4.jpg&width=1248&dpr=4&quality=100&sign=ebd383ce&sv=2) [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdo-yzgaido#hajimeni) **はじめに** ----------------------------------------------------------------------------------------------------------- 1. 質問を入力欄に入力します。 2. Oracle Mode を有効にします。 3. 「開始」 をクリックします。 4. プランナーを確認し、「確認して実行」 を選択して実行します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FjKL3HRJmEL8iF3vwSTPC%2FOracle-%25E4%25BD%25BF%25E7%2594%25A8%25E6%258C%2587%25E5%258D%2597-%25E5%25BC%2580%25E5%25A7%258B.gif&width=768&dpr=4&quality=100&sign=abc8f49f&sv=2) 「ORACLE MODE」 を有効にしたままにする [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdo-yzgaido#recipe-nowosuruniha) Recipe の実行進捗を確認するには --------------------------------------------------------------------------------------------------------------------------------- 1. ノードを選択します。 2. ページ下部の入力フィールドにカーソルを移動させます。 3. recipe の実行進捗が表示されます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2F1dCYCFst5Sca1Ugd1PEM%2FOracle-%25E4%25BD%25BF%25E7%2594%25A8%25E6%258C%2587%25E5%258D%2597-%25E6%259F%25A5%25E7%259C%258Brecipe%25E7%259A%2584%25E6%2589%25A7%25E8%25A1%258C%25E8%25BF%259B%25E5%25BA%25A6.gif&width=768&dpr=4&quality=100&sign=b6c95ba7&sv=2) 各ステップの前にある円はその現在のステータスを示しています [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdo-yzgaido#norecipewosuru) **過去のRecipeを再実行する方法** ------------------------------------------------------------------------------------------------------------------------------ 1. 確認したい recipe のノードをクリックします。 2. ダイアログボックスにカーソルを合わせると、自動的に recipe が表示されます。 3. 「Re-run」 をクリックして確認ページを開きます。 4. 「確認して実行」をクリックすると、プロセスが開始されます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fob2PVTif5kcTSCFjDg4u%2FOracle-%25E4%25BD%25BF%25E7%2594%25A8%25E6%258C%2587%25E5%258D%2597-%25E9%2587%258D%25E5%25A4%258D%25E6%2589%25A7%25E8%25A1%258C%25E4%25B8%258A%25E4%25B8%2580%25E4%25B8%25AArecipe.gif&width=768&dpr=4&quality=100&sign=ef528b95&sv=2) プロセスステップのステータス表示 [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdo-yzgaido#jikyanbasunishiiburanchiwosuru) **同じキャンバス内に新しいブランチを作成する方法** ---------------------------------------------------------------------------------------------------------------------------------------------------- 1. 「新しいプランチ+」 ボタンをクリックします。 2. ページ下部に表示されるダイアログボックスに、新しいクエリまたはトピックを入力します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2F7iCuoE2BGVJyCsTIQCaN%2FOracle-%25E4%25BD%25BF%25E7%2594%25A8%25E6%258C%2587%25E5%258D%2597-%25E5%25BC%2580%25E5%2590%25AF%25E6%2596%25B0%25E5%2588%2586%25E6%2594%25AF.gif&width=768&dpr=4&quality=100&sign=48d997de&sv=2) 1つのキャンバスで複数のブランチを使用する方法 [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdo-yzgaido#recipe-tenpurtonoi) Recipe **テンプレートの使い方** ---------------------------------------------------------------------------------------------------------------------------------- 1. 「探索」 セクションで希望するテンプレートのキーワードを検索します。 2. 「Use this recipe」 をクリックして詳細を表示します。 3. 「確認して実行」 をクリックしてテンプレートを適用します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FD0JALV2zTHkhmbr2C8JB%2FOracle-%25E4%25BD%25BF%25E7%2594%25A8%25E6%258C%2587%25E5%258D%2597-%25E5%25A6%2582%25E4%25BD%2595%25E4%25BD%25BF%25E7%2594%25A8%25E6%25A8%25A1%25E7%2589%2588.gif&width=768&dpr=4&quality=100&sign=88b3c8ff&sv=2) コミュニティ内で選択可能な専門的なテンプレートをご用意しています [PreviousOracle モードはどのような問題解決に適していますか？](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdohadonoyounanishiteimasuka) [NextOracle モードの高度なテクニック](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-mdononatekunikku) Last updated 10 months ago Was this helpful? --- # プロンプトエンジニアリングの基礎 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fg6n2Kda04BzP3TPTHtPJ%2F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE4.jpg&width=1248&dpr=4&quality=100&sign=ebd383ce&sv=2) 生成 AI は非常に有用ですが、人間の指示が必要です。多くの点で、生成 AI は企業の新しい有能なインターンのようなものだと言えます。大きな可能性を秘めていますが、最適なパフォーマンスを発揮するには明確な指示が必要です。生成 AI を適切に指示する能力は、強力なスキルです。生成 AI を指示するには、通常テキストベースの指示であるプロンプトを送信します。プロンプトは AI アシスタントに提供される入力であり、出力に大きな影響を与えます。効果的なプロンプトは、構造化され、明確で、簡潔かつ目的を持ったものである必要があります。 **適切に構造化されたプロンプトの作成方法** 構造化されたプロンプトとは、明確な論理と構成で組み立てられたものを指します。例えば、AI モデルに記事を生成させたい場合、プロンプトには以下のような情報を含める必要があるかもしれません： * 記事のメインテーマ * 記事の概要 * 希望する文体 基本的なディスカッションの質問の例を見てみましょう： > 「地球が直面している最も緊急性の高い環境問題とは何か、そしてこれらの問題に対処するために個人ができる取り組みは何か？」 これを簡単なアシスタント用のプロンプトに変換するには、文頭に 「以下の質問に答えてください：」 と追加するだけで済みます。 Copy 以下の質問にお答えください： 地球が直面している最も緊急の環境問題は何か、そしてこれらの問題に対処するために個人ができる対策は何かについて説明してください。 このプロンプトから生成される結果は一貫性に欠け、1〜2文程度の回答もあるため、理想的とは言えません。通常、議論の回答は複数の段落で構成されるべきです。効果的なプロンプトは、フォーマットと内容に関する具体的な指示を提供する必要があります。一貫性と品質を向上させるには、言語の曖昧さを排除する必要があります。 以下に、改善されたプロンプトを示します： Copy 以下の質問に対して、序論、本論、結論を含む詳細なエッセイを書いてください： 地球が直面している最も緊急の環境問題は何ですか？ これらの問題に対処するために個人ができる対策は何ですか？ 2番目のプロンプトでは、より長く、構造が改善された出力が生成されました。プロンプトで「エッセイ」という用語を意図的に使用したのは、AI アシスタントがエッセイの定義を理解しているため、より一貫性のある構造化された応答を生成する可能性が高いからです。 #### [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/puronputoenjiniaringuno#towosaseru) 品質と効果を向上させる方法 プロンプトの品質と効果を高めるための主要な方法がいくつかあります： * 要件を明確に定義する：AI モデルは可能な限りあなたのニーズに応えようとします。要件があいまいな場合、出力が期待に沿わない可能性があります。 * 正しい文法とスペルを使用する：モデルはあなたの言語スタイルを模倣する傾向があります。そのため、あなたの言語に問題がある場合、出力にもその問題が反映される可能性があります。 * 十分なコンテキストを提供する：モデルは、あなたが提供するコンテキスト情報に基づいて出力を生成します。コンテキストが不十分な場合、望ましい結果を得ることが難しくなる可能性があります。 効果的なプロンプトを作成した後は、生成された結果を洗練させる必要があります。これには、単語数制限などの特定の制約に合わせて出力を調整したり、異なる生成結果から概念を組み合わせたりすることが含まれます。 反復的な改善の簡単な方法は、複数の出力を生成し、それらを確認して使用されている概念や構造を理解することです。出力を評価した後、最も適切なものを選択し、一貫性のある回答にまとめることができます。もう一つの反復的なアプローチは、小さく始めて徐々に拡張していくことです。これには複数のプロンプトが必要です：最初のプロンプトで1〜2段落を書き、その後、既に書かれた内容を拡張するための追加プロンプトを使用します。 以下は、哲学的な議論の題材となる可能性のある質問です： > 「数学は発明されたものか、それとも発見されたものか？慎重な推論を用いて答えを説明してください。」 この質問を簡単なプロンプトに追加すると、次のようになります： Copy 以下の質問に答える導入段落を書いてください： 数学は発明されたものか、それとも発見されたものか？慎重な推論を用いて答えを説明してください。 これらの考え方について、私はいくつかの回答を生成し、その中から気に入ったものを選びました。 Copy 数学が発明されたものか、それとも発見されたものかという問題は、哲学者や数学者の間で長年議論の的となってきました。 一部の人々は、数学の概念や原理は人間の思考とは独立して存在するため、発明というよりも発見されたものと考えることができると主張しています。 例えば、数、比率、幾何学的形状といった概念は自然界で観察され、体系的な研究へと抽象化することができます。 これは良い出発点です。その後、別のプロンプトを使用して内容を拡張しました：（下面有プロンプト） Copy 以下の《プロンプト》に対する詳細な短文エッセイを執筆しています： 数学は発明されたものか、発見されたものか？慎重な論理を用いて回答を説明してください。 現在の進捗状況は以下の通りです： 数学が発明されたものか発見されたものかについては、哲学者や数学者の間で長く議論されてきました。 一部の人々は、数学の概念や原理は人間の思考とは独立して存在するため、発明というよりも発見されたものと考えられると主張しています。 例えば、数、比率、幾何学的形状などの概念は自然界で観察され、体系的に抽象化し研究することができます。 エッセイの次の段落を執筆してください。 私はこのプロンプトを使用して複数の展開を生成し、気に入ったものを選択しました。最終的に以下のような結果となりました： Copy 哲学者と数学者の間で議論が続いています： 数学は発明されたのか、それとも発見されたのか？ ある人々は、数学的概念や原理は人間の思考とは独立して存在するため、発明というよりも発見と考えることができると主張します。 例えば： 数、比率、幾何学的形状などの概念は自然界で観察され、体系的に抽象化して研究することができます。 一方で、数学は主に人間の心の発明であると主張する人々もいます。 彼らの主張には以下のようなものがあります： 数学の原理は、人間の思考の特徴である論理的推論に基づいています。 例えば、ユークリッド幾何学は、証明なしに真として受け入れられる抽象的な公理と前提に基づいています。 したがって、幾何学は発見というよりも人間の思考の発明とみなすことができます。 同様に、物理現象をモデル化し予測するために使用される数学的公式や方程式も人間の推論の産物です。 拡張プロンプトを使用することで、段階的に文章を作成し、各ステップを反復的に改善することができます。このアプローチは、より高品質な出力が求められる場合や、段階的な修正が望ましいシナリオで特に有効です。 [Previous高度な機能 — ノード参照](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/na-ndo) [Nextflowith ナレッジガーデンの概要](https://doc.flowith.io/flowith-heyoukoso/knowledge-garden-narejjigden/flowith-narejjigdenno) Last updated 10 months ago Was this helpful? --- # 流式聊天使用指南 | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) [](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-shi-yong-zhi-nan#kai-shi-chuang-jian) 开始创建 ------------------------------------------------------------------------------------------------------------- 点击 「新对话」 后： 1. 在对话框中输入问题，注意要保持 Oracle 模式关闭 2. 如果需要自定义模型，可在左下方选择模型 3. 点击 Start 按钮 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FxSwA7sX4kAhRjsu0fFBk%252F9.11%25E5%2592%258C9.9%25E5%2593%25AA%25E4%25B8%25AA%25E5%25A4%25A7.gif%3Falt%3Dmedia%26token%3Dff9f545b-5854-4259-a96e-7ccbfe6f177b&width=768&dpr=4&quality=100&sign=96a8e3b9&sv=2) 开启新对话 [](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-shi-yong-zhi-nan#dui-ai-gei-chu-de-hui-da-ji-xu-zhui-wen) 对 AI 给出的回答继续追问 ------------------------------------------------------------------------------------------------------------------------------------------- 1. 点击想追问的节点 2. 在对话框中输入你想提问的问题 3. 按下 Enter 键 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FpT2fTgC3wqaSQi2tK8Dw%252F%25E8%25BF%25BD%25E9%2597%25AE.gif%3Falt%3Dmedia%26token%3Dcb79805c-94e5-4e67-ae51-8bbbec578c63&width=768&dpr=4&quality=100&sign=4451366d&sv=2) 对 AI 的回答可以进行进一步追问 [](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-shi-yong-zhi-nan#chong-xin-sheng-cheng-te-ding-jie-dian-zhong-de-nei-rong) 重新生成特定节点中的内容 ---------------------------------------------------------------------------------------------------------------------------------------------------------- 对节点中的内容不满意时，点击重新运行节点，即可一键生成新的内容 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FAdJJyqAxtw8L4yikDYfh%252F%25E9%2587%258D%25E6%2596%25B0%25E8%25BF%2590%25E8%25A1%258C.gif%3Falt%3Dmedia%26token%3D50892876-7f70-4291-9282-c79284304db1&width=768&dpr=4&quality=100&sign=dd696a8&sv=2) 一键重新运行该节点 [](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-shi-yong-zhi-nan#zhan-kai-jie-dian-zhong-de-nei-rong) 展开节点中的内容 --------------------------------------------------------------------------------------------------------------------------------- 如果节点中的内容过多不便于展示， 1. 选中节点 2. 点击 「展开节点」 即可 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FIlNt5ZVlA0rplwSCZ8js%252F%25E5%25B1%2595%25E5%25BC%2580.gif%3Falt%3Dmedia%26token%3Df1c3726b-fa8d-4cc0-bbc4-53f388144a1b&width=768&dpr=4&quality=100&sign=90ee78fe&sv=2) 展开节点查看完整内容 [](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-shi-yong-zhi-nan#fu-zhi-jie-dian-zhong-de-jie-guo) 复制节点中的结果 ------------------------------------------------------------------------------------------------------------------------------ 1. 选中节点 2. 点击 「复制节点」 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252F3DfE0GifEEk85bd27k9o%252F%25E5%25A4%258D%25E5%2588%25B6%25E8%258A%2582%25E7%2582%25B9.gif%3Falt%3Dmedia%26token%3Dd8033ff2-713a-441f-b320-b182f264259a&width=768&dpr=4&quality=100&sign=61cc96a9&sv=2) 对生成的内容一键复制 [](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-shi-yong-zhi-nan#shan-chu-jie-dian) 删除节点 ----------------------------------------------------------------------------------------------------------- **删除单个节点** 1. 选中节点 2. 点击 「删除节点」 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FxCUkZEktplqP1p04rniv%252F%25E5%2588%25A0%25E9%2599%25A4%25E8%258A%2582%25E7%2582%25B9.gif%3Falt%3Dmedia%26token%3D73e02308-e41e-4053-9a73-34737ce2c8cb&width=768&dpr=4&quality=100&sign=aa8a9df1&sv=2) 一键删除节点 **删除多个节点** 1. 按住 「shift」 键，依次点击想要删除的多个节点 2. 按下 「backspace」 键 3. 确认删除 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FuY4dqDZEMpTxP35JUbwv%252F%25E5%2588%25A0%25E9%2599%25A4%25E5%25A4%259A%25E4%25B8%25AA%25E8%258A%2582%25E7%2582%25B9.gif%3Falt%3Dmedia%26token%3D168bf960-cece-4c6f-8876-d386e9857ce8&width=768&dpr=4&quality=100&sign=b8e78002&sv=2) 需要长按 「shift」 键 [](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-shi-yong-zhi-nan#fen-xiang-wo-de-flow-gei-peng-you) 分享我的 flow 给朋友 ------------------------------------------------------------------------------------------------------------------------------------ 1. 点击页面左上角的 Share 2. 选择权限 「可查看」 3. 复制链接并分享 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252Fob84pkgUw3g2EqrqCAjI%252F%25E5%2588%2586%25E4%25BA%25AB%25E7%25BB%2599%25E6%259C%258B%25E5%258F%258B.gif%3Falt%3Dmedia%26token%3D595a185e-253c-4895-b4bd-adc73dd27a87&width=768&dpr=4&quality=100&sign=d1d1ac54&sv=2) 邀请朋友欣赏你的创意 [](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-shi-yong-zhi-nan#yao-qing-peng-you-gong-tong-chuang-zuo) 邀请朋友共同创作 ------------------------------------------------------------------------------------------------------------------------------------ 1. 点击页面左上角的 Share 2. 选择权限 「可编辑」 3. 复制链接并分享 4. 使用评论功能在画布中进行交流 1. 点击页面左上角 「插入评论」 2. 点击你想要的评论的节点 3. 再次点击页面左上角的 「插入评论」 退出 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252FcEBs1PXmi8j89x0hEe74%252F%25E5%2585%25B1%25E5%2588%259B1.gif%3Falt%3Dmedia%26token%3Df282a7c7-1843-41d8-9dba-62afc6f6caa7&width=768&dpr=4&quality=100&sign=a00df74&sv=2) 开启编辑权限并分享 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3657150132-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FAVkWkYQheefIjsWOM7id%252Fuploads%252F3aMhNnC2z3dvaGmAzxhH%252F%25E6%258F%2592%25E5%2585%25A5%25E8%25AF%2584%25E8%25AE%25BA.gif%3Falt%3Dmedia%26token%3D76d0bb0b-db35-49fb-aaa1-24d90107de34&width=768&dpr=4&quality=100&sign=ddbc8a2&sv=2) 点击节点并评论 [Previous流式聊天模式介绍](https://doc.flowith.io/cn/flowith-normal/liu-shi-liao-tian-mo-shi-jie-shao) [Next高级玩法 — 节点引用](https://doc.flowith.io/cn/flowith-normal/gao-ji-wan-fa-jie-dian-yin-yong) Last updated 11 months ago Was this helpful? --- # Canvas Navigation in Agent Mode | Flowith Official Document [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#canvas-execution-prompt-to-plan) Canvas Execution: Prompt to Plan ----------------------------------------------------------------------------------------------------------------------------------------------- ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#id-1.-prompt-parsing-and-seed-discovery) 1\. Prompt Parsing and Seed Discovery * **Trigger:** User enters a prompt and activates Neo in Agent Mode. * **Process:** Neo analyzes the prompt for **key terms**, including: * Named entities (e.g., **“May 2025 Sales Report”**, **“Q2 Consumer Behavior Summary”**, **“Chimera GPU Whitepaper”**) * Goal-based language (e.g., “summarize”, “generate”, “compare”) > This parsing is automatic. No special formatting is needed by the user — Neo extracts intent and targets naturally from conversational input. * **Knowledgebase Usage (if enabled):** Neo queries the **Knowledge Base** for any contextually matching documents. These documents are converted into “**seeds**” — key fragments of relevant data — which are used to **grow a context-aware Recipe**. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252F49KhLtryS4WaaDR70oTR%252FKowledge_Base_Load.gif%3Falt%3Dmedia%26token%3D0ed86a2f-fb32-46e6-9fc6-77de7fb919ff&width=768&dpr=4&quality=100&sign=5a2d677c&sv=2) Agent Neo Scanning the enabled Knowledge Base ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#id-2.-knowledge-garden-and-recipe-generation) 2\. Knowledge Garden and Recipe Generation * **Concept:** Neo builds a **Recipe**, a living execution plan made up of evolving subtasks. * **With Knowledge Base Enabled:** * Extracted seeds are used to **enrich** the Recipe with domain-specific context. * Seeds are re-quoted inside the canvas so you can trace what context was used. * **Without Knowledge Base:** * Neo builds a base-level Recipe using model knowledge, tools, and live search. * **Additional Controls:** * You can **pause** an active Recipe at any time to temporarily halt execution. * You can also **duplicate** a Recipe to fork it into a new session for experimentation or comparison. ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#id-3.-task-planning-and-tool-allocation) 3\. Task Planning and Tool Allocation Once the Recipe is seeded: * **Neo activates the Dispatcher.** This allocator breaks down the Recipe into actionable subtasks and selects tools for each step. * **Toolset Sampling:** Based on your goal, Neo dynamically selects from tools such as: * Web Search * Code Generator * File Uploader * Chart Interpreter * Terminal (if enabled) * **Live Plan Rendering:** The Recipe and subtasks are visualized directly in the canvas. Each task node can be expanded to see: * Chosen Tool * Reasoning * Output * Follow-up tasks (auto-generated) > If a task fails, Neo re-routes. It will adapt the Recipe and try alternative tools or updated phrasing — even mid-plan. The remaining steps to complete the task is estimated by Neo and displayed at the bottom of the recipe list ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FbdDreUnNJd7M1Ak6B8ME%252FRecipe_Execution.gif%3Falt%3Dmedia%26token%3D93d323cd-35cb-456c-a38d-3699adbe4058&width=768&dpr=4&quality=100&sign=b36cca9b&sv=2) ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#id-4.-execution-begins) 4\. Execution Begins After the plan is finalized, the **Start** button launches execution. * You can pause, stop, or adjust task inputs live. > Neo continues refining the Recipe in real time, responding to results, user edits, and discovered context. This dynamic execution lets Neo outperform static agents that cannot re-plan. [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#common-node-types-in-neos-canvas) Common Node Types in Neo's Canvas ------------------------------------------------------------------------------------------------------------------------------------------------- As Neo executes a task, it dynamically generates **node blocks** inside the canvas. Each node represents a specific action or tool usage within the current Recipe. Understanding these common node types helps you trace Neo’s reasoning and interact more effectively with ongoing or completed plans. ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#id-1.-gen-document-node) 1\. Gen Document Node **Purpose:** Neo uses this node to draft written plans such as **roadmaps**, **task briefs**, and **multi-step strategies** based on the current Recipe context. * **Features:** * Click the **down arrow** icon to expand or collapse the full document view (appears on the right side of the screen). * Documents are version-aware and update based on upstream changes in the Recipe. > Tip: Review the Documents to provide more concise follow-ups with neo about edits. You can also Right Click this node to directly add the contents to your Knowledge Base ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FM9k8D68wd2tdjoljwinJ%252FGen_Document.gif%3Falt%3Dmedia%26token%3D43e18fc6-cb1f-4b4b-920d-dd5aab6553b4&width=768&dpr=4&quality=100&sign=9c25e367&sv=2) Click the Down Arrow to expand contents on the right of screen ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#id-2.-online-search-node) 2\. Online Search Node **Purpose:** Triggers a deep search across multiple platforms to gather external data necessary for informed task planning. * **Features:** * Displays links and excerpts from search results. * Includes a **source toggle** to reveal or hide exact URLs and timestamps. * Neo may create multiple Online Search nodes per task if additional data is required. > Example: Searching for “Chimera GPU specs” results in this node, which then influences downstream coding or writing steps. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FR9A2wgV3bpLI62d0h8Zb%252FSearch_Tool_Node.gif%3Falt%3Dmedia%26token%3D90a30935-02f7-4efc-bfc0-9c777a82dfa6&width=768&dpr=4&quality=100&sign=247f4be&sv=2) Click the Toggle Button below the Node to view sources ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#id-3.-gen-webpage-node) 3\. Gen Webpage Node **Purpose:** This is Neo’s **code generation node**, typically used for writing websites, frontends, scripts, or downloadable source files. You can click the Down Arrow icon to review the code ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252Fc4F4IXPlcCk5OWmSjpGk%252Fgen_webpage.gif%3Falt%3Dmedia%26token%3Db54e3493-d490-49ed-ad46-0557b82b5322&width=768&dpr=4&quality=100&sign=5a0f70a3&sv=2) Click the Toggle Button to review Source Code similar to Gen Document Node **Features:** * Refine with Neo: Prompt Neo to edit anything rendered inside the node * Edit in Fullscreen: Opens an inline fullscreen editor ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252F4M0ARJlBoFPKQ3AvunYr%252Fedit_webpage.gif%3Falt%3Dmedia%26token%3D4c3d807d-56e7-482e-8f92-ee2c695a1c5a&width=768&dpr=4&quality=100&sign=50d21bfc&sv=2) Easily Reposition and Edit Elements * Open in New Tab: Launches the code output in a sandboxed preview * Download: Saves the raw code as a ZIP file > This node is used extensively for technical recipes involving HTML, JS, CSS, Markdown, or static site builds. ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#id-4.-all-task-coimplete-node) 4\. All Task Coimplete Node **Purpose:** Signals that Neo has finished executing all Recipe steps. * **Features:** * Shows a **completion banner** inside the canvas. * Contains a **follow-up input dialog** so you can extend or refine the task without resetting the entire canvas. > You can treat this as a checkpoint — review the output, save results, or continue iterating. ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#id-5.-ask-user-more-node) 5\. Ask User More Node **Purpose:** Activated automatically when Neo detects that the user’s prompt is too vague, or context is insufficient to proceed. * **Features:** * Contains a **clarification prompt** that Neo fills out. * Accepts direct text input and supports file uploads using the **paperclip icon**. * Can also be **manually triggered** by the user or agent for branching logic. > This node helps ensure clarity, especially when multi-step tasks require exact requirements before continuing. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FsNLRU7640iYjfe1uB8tw%252Fask_user_more.gif%3Falt%3Dmedia%26token%3D047aae0b-7b56-4be7-8dbe-92d9a8abfdee&width=768&dpr=4&quality=100&sign=11ee560c&sv=2) You can call the Ask User More tool by a simple prompt [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#recipe-nodes-and-task-control) Recipe Nodes & Task Control ---------------------------------------------------------------------------------------------------------------------------------------- ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#what-is-a-recipe-node) What is a Recipe Node? Recipe Nodes are the **elliptical step indicators** that appear between each major task block in the canvas. They represent the logical transition from one subtask to the next in Neo's active Recipe. These nodes visually mark where Neo has paused to evaluate a result, shift context, or prepare for the next operation. ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#functionality) Functionality **Restart from this point:** Hover over the node and click the **Refresh icon** to reinitialize Neo’s planning from this specific step in the Recipe. * **Contextual Rebuild:** When restarted, Neo regenerates the Recipe forward from this node using the latest outputs, edits, or external file uploads, without needing to rerun previous steps. > This is useful for correcting logic paths or introducing new information without discarding prior work. ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#when-to-use-it) When to use it * You want to re-test from a specific step onward * You need to fix a failed task or broken output without starting over ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FN5MX6JwCrG1SqQmeksYV%252Frestart_task.gif%3Falt%3Dmedia%26token%3Db828ecff-2cde-43a4-9ceb-a0e5b7355314&width=768&dpr=4&quality=100&sign=fa93be8a&sv=2) Restart a Task from any Recipe Node > Neo’s dynamic planning engine will respect the prior flow while adapting downstream logic, preserving the fluidity of the Recipe. [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#downloading-agent-files) Downloading Agent Files ------------------------------------------------------------------------------------------------------------------------------ ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#accessing-the-file-panel) Accessing the File Panel After Neo completes any file-based task — such as generating documents, webpages, or analysis — all resulting files are stored under the **Files panel** in the left sidebar. * **Click the** `**Files**` **button** (folder icon) to open the list of all downloadable items created during the current Recipe. ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#download-options) Download Options Once inside the Files panel, you have two options for retrieving Neo’s output: ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#id-1.-manual-review-and-download) 1\. Manual Review and Download * **Browse the file tree** to see all available items by title. * Each file is labeled with: * A unique name based on the content or task * A download count indicator (how many times it was revised or regenerated) * Click on any filename to **download it individually**. > Use this method when you want to preview or selectively download results. ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#id-2.-download-all-files-at-once) 2\. Download All Files at Once * Click the **Download All Files** icon (⇩) at the top-right of the Files panel. * This will **bundle all generated files into a single** `**.zip**` **archive** and initiate an immediate download. > Ideal for exporting full project outputs, such as multiple reports, bundled HTML+CSS+JS files, or full document packs. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FUgoYbWL9Om9iB0fvhi17%252Fdownload_files.gif%3Falt%3Dmedia%26token%3D8ffccbb9-95ee-4d12-9033-0e581b6ddbed&width=768&dpr=4&quality=100&sign=9cb7885d&sv=2) Download All Files Button Located Beside "Files generated by the Agent" text [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#how-to-create-new-branches-in-the-same-canvas) How to Create New Branches in the Same Canvas -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- * Click on the "New Branch" button. * In the dialog box that appears at the bottom of the page, enter your new query or topic. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FUZSier34VM9mXed3xp2P%252FOracle-%25E4%25BD%25BF%25E7%2594%25A8%25E6%258C%2587%25E5%258D%2597-%25E5%25BC%2580%25E5%2590%25AF%25E6%2596%25B0%25E5%2588%2586%25E6%2594%25AF-new-branch.gif%3Falt%3Dmedia%26token%3Dde99c89c-a8a5-4abe-bfbe-afb21e499fcf&width=768&dpr=4&quality=100&sign=b7e83e24&sv=2) You Can Host Multiple Branches in the Same Flow [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#sharing-a-flow) Sharing a Flow ------------------------------------------------------------------------------------------------------------ ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#using-the-share-button) Using the Share Button To collaborate on or showcase your current Flow, use the `**Share**` **button** located at the top-left of the Canvas. * Click the `Share` button to generate a **unique Flow URL**. * This link allows others to view or join the session. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252Fp85EY2lXUa7YedocuJbi%252Fshare_button.gif%3Falt%3Dmedia%26token%3Dbbac9ac4-d4b4-456c-ad0e-5a4e7d39be31&width=768&dpr=4&quality=100&sign=92b8ee69&sv=2) Only Grant Edit Permissions to Users You Trust ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#important-sharing-rules) Important Sharing Rules * **Default Share Mode:** By default, sharing a Flow will **only grant view access to replies** — the recipient can see what Neo generated, but **cannot edit or interact** with the Recipe. * **Granting Edit Access:** After someone has joined using the shared link, you can **manually assign edit permissions** from the member management dropdown. Edit permissions must be granted for a user to view the contents of a Flow, or else they will only see the replay of the Flow > Only collaborators with edit access can modify task inputs, rerun nodes, or inject new files into the session. ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#swapping-to-regular-mode) Swapping to Regular Mode If you don’t need full Agent execution, you can switch to **Regular Mode** to prompt a specific model directly. This is ideal for fast responses, brainstorming, and comparisons ### [](https://doc.flowith.io/agent-mode-neo/canvas-navigation-in-agent-mode#available-modes-in-regular-mode) Available Modes in Regular Mode **Regular (Prompt a Model)** Choose from available models (e.g., Claude, GPT, Gemini) and ask questions or generate text without triggering Agent Mode. * **Online Search** Send your prompt to a model with access to real-time web search. Choose which model will handle the query + summarization. * **Image / Video** Prompt any supported vision model for analysis, captioning, or enhancement suggestions. * **Prompt Enhance** Automatically rewrites your prompt to be more detailed, structured, or effective . * **Comparison Mode** Run the same prompt across multiple models side-by-side and compare their outputs in a unified view. * * * > Regular Mode is flexible, fast, and model-specific. Use it when you want direct control over how a prompt is handled—without planning or execution steps. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2F3690352926-files.gitbook.io%2F%7E%2Ffiles%2Fv0%2Fb%2Fgitbook-x-prod.appspot.com%2Fo%2Fspaces%252FjTYyhELhBzd4Ov8Nl3WG%252Fuploads%252FrIPEFMBRMmeA8KGtsQrY%252Fregular_mode_toggle.gif%3Falt%3Dmedia%26token%3D8f5078b0-96c9-43ca-8051-38c254c26cbd&width=768&dpr=4&quality=100&sign=9be15671&sv=2) [PreviousAgent Mode User Guide](https://doc.flowith.io/agent-mode-neo/agent-mode-user-guide) [NextAgent Neo Use Cases](https://doc.flowith.io/agent-mode-neo/agent-neo-use-cases) Last updated 1 month ago Was this helpful? --- # Flow モード ユーザーガイド | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fstatic-2v.gitbook.com%2F_next%2Fstatic%2Fmedia%2Fdefault-page-cover.a2ccd7e9.svg&width=1248&dpr=4&quality=100&sign=899888f5&sv=2) [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdo-yzgaido#hajimeni) **はじめに** ------------------------------------------------------------------------------------------------- 「新しいチヤツト」 をクリックした後： 1. チャットボックスに質問を入力します。Oracle モードがオフになっていることを確認してください。 2. モデルをカスタマイズする必要がある場合は、左下のオプションから選択できます。 3. 「Start」 をクリックして会話を始めます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FDmdcJpNh7zCZ5Kcu6FT6%2F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%25BC%2580%25E5%25A7%258B%25E5%2588%259B%25E5%25BB%25BA.gif&width=768&dpr=4&quality=100&sign=a7d5456c&sv=2) 新しい会話を始める [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdo-yzgaido#ai-nonisuru) AI の回答に関する追加質問 ---------------------------------------------------------------------------------------------------------- 1. さらに詳しく尋ねたいノードをクリックします。 2. ダイアログボックスに追加の質問を入力します。 3. Enter キーを押します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2F0zDJEl5tslTQkDzqV9DO%2F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E7%25BB%25A7%25E7%25BB%25AD%25E8%25BF%25BD%25E9%2597%25AE.gif&width=768&dpr=4&quality=100&sign=9026bb1b&sv=2) AI の応答に対するフォローアップ質問 [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdo-yzgaido#nondonokontentsuwosuru) 特定のノードのコンテンツを再生成する ------------------------------------------------------------------------------------------------------------------------- ノード内のコンテンツに満足できない場合は、「再実行」 ボタンをクリックするだけで、ワンクリックで新しいコンテンツを瞬時に生成することができます。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fy2Rh8FY9ML77VT6gJ10u%2F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E9%2587%258D%25E6%2596%25B0%25E7%2594%259F%25E6%2588%2590%25E8%258A%2582%25E7%2582%25B9%25E5%2586%2585%25E5%25AE%25B9.gif&width=768&dpr=4&quality=100&sign=2ddeccf0&sv=2) ワンクリックでノードを再実行 [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdo-yzgaido#ndonowo) **ノードの内容を展開** ----------------------------------------------------------------------------------------------------- ノードに表示しきれないほど多くの情報が含まれている場合： 1. ノードを選択します。 2. 「もっと見る」 をクリックします。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FhdiCt3LJbFS1UuPsF8uc%2F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%25B1%2595%25E5%25BC%2580%25E8%258A%2582%25E7%2582%25B9%25E5%2586%2585%25E5%25AE%25B9.gif&width=768&dpr=4&quality=100&sign=307494e&sv=2) ノードを展開してすべての内容を表示 [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdo-yzgaido#ndokarawokopsuru) **ノードから結果をコピーする** ------------------------------------------------------------------------------------------------------------------ 1. ノードを選択します。 2. 「コピー」 をクリックします。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FJv6LLNIMP532ZpcBUKsd%2F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%25A4%258D%25E5%2588%25B6%25E7%25BB%2593%25E6%259E%259C.gif&width=768&dpr=4&quality=100&sign=c403d8c1&sv=2) 生成されたコンテンツをワンクリックでコピー [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdo-yzgaido#ndono) **ノードの削除** ------------------------------------------------------------------------------------------------ **単一のノードを削除する** 1. ノードを選択します。 2. 「削除」 をクリックします。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2F2uw7SCPJcrViYBru6aEc%2F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%2588%25A0%25E9%2599%25A4%25E5%258D%2595%25E4%25B8%25AA%25E8%258A%2582%25E7%2582%25B9.gif&width=768&dpr=4&quality=100&sign=3a4d18c7&sv=2) ノードの素早い削除 **複数のノードを削除する方法** 1. 「Shift」 キーを押しながら、削除したいノードを一つずつクリックします。 2. 「Backspace」 キーを押します。 3. 削除の確認を行います。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FUTW28ofppmSSjuaPMQMX%2F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%2588%25A0%25E9%2599%25A4%25E5%25A4%259A%25E4%25B8%25AA%25E8%258A%2582%25E7%2582%25B9.gif&width=768&dpr=4&quality=100&sign=d3349bc7&sv=2) 「Shift」 キーを押し続ける [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdo-yzgaido#to-flow-wosuru) 友達と flow を共有する方法 --------------------------------------------------------------------------------------------------------------- 1. ページ左上の 「共有」 をクリックします。 2. 権限を 「閲覧可能」 に設定します。 3. リンクをコピーして共有します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FHubmQld7ZviMhhMkccDQ%2F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E5%2588%2586%25E4%25BA%25AB.gif&width=768&dpr=4&quality=100&sign=ca9e7b4f&sv=2) 友達を招待して、あなたの創造性を共有しよう [](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdo-yzgaido#tosuru) **友達と協力する** -------------------------------------------------------------------------------------------------- 1. ページ左上の 「共有」 ボタンをクリックします。 2. 権限を 「編集可能」 に設定します。 3. リンクをコピーして共有します。 4. キャンバス内でのコミュニケーションにはコメント機能を使用します。 1. ページ左上の 「コメントを追加」 をクリックします。 2. コメントしたいノードをクリックします。 3. 再度左上の 「コメントを追加」 をクリックし、コメントモードを終了します。 ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FztUebHpKvH7c45U3O0MQ%2F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E9%2582%2580%25E5%258F%258B%25E5%2585%25B1%25E5%2588%259B-%25E7%25BC%2596%25E8%25BE%2591.gif&width=768&dpr=4&quality=100&sign=b48a7182&sv=2) 編集を有効にして共有する ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2F1jucJPUFKosyKetODJA5%2F%25E6%25B5%2581%25E5%25BC%258F%25E6%258C%2587%25E5%258D%2597-%25E9%2582%2580%25E5%258F%258B%25E5%2585%25B1%25E5%2588%259B-%25E8%25AF%2584%25E8%25AE%25BA.gif&width=768&dpr=4&quality=100&sign=946fe9ca&sv=2) 特定のノードをクリックして的確なコメントを追加する [PreviousFlow モードの紹介](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/flow-mdono) [Next高度な機能 — ノード参照](https://doc.flowith.io/flowith-heyoukoso/flow-mdo-mdo/na-ndo) Last updated 10 months ago Was this helpful? --- # Current Capabilities Supported by Oracle | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2FjTYyhELhBzd4Ov8Nl3WG%2Fblobs%2FFe0YxPmfFaclcndbPzUc%2FAmazing%2520Mockups.jpeg&width=1248&dpr=4&quality=100&sign=e138a96b&sv=2) The following is a list of external tools currently supported by the Flowith Oracle mode. This list is continuously being updated. If you have specific tools you'd like Oracle to support, or if you're interested in integrating your product's API with Oracle, we welcome you to contact us! **Tool Name** **Tool Function** analyze\_website\_traffic analyze specific website traffic; get website metrics; check website engagement ask\_user\_more ask user to input more information deep\_thinking deep analysis on a subject gen\_webpage generate web page, create website, design user interface, build online platform, develop web application, construct digital presence, craft internet portal, produce web-based solution, create ppt presentations, design infographics, generate charts and graphs, visualize data, create mindmaps and flowcharts. generate\_image generate images; draw images; text to image generations; draw a cute cat generate\_python help user run python code,code interpreter; do complex mathematical operations; do data visualization; run python code online\_search search realtime information by Google read\_website read a website's content; scrape website information; retrieve online articles search\_google\_scholar search academic papers; find scholarly articles; search\_images google image search; search\_jobs search for jobs search\_linkedin\_company search company on LinkedIn search\_linkedin\_people search people on LinkedIn, find LinkedIn profiles search\_patents search patents, find patent information, lookup inventions search\_products search for products, find items to buy, compare prices, shopping information search\_reddit\_posts search Reddit posts search\_stackexchange\_question search stackoverflow posts search\_twitter search tweets, find twitter posts search\_videos search YouTube videos, video search search\_websites search for websites, find links, google search search\_xiaohongshu\_notes search xiaohongshu posts send\_email send email, send mail, g-mail, contact a person by e-mail send\_rapid\_email send email; fast email delivery summarize\_and\_decide summarize contents translate\_text translate content user provided Last updated 10 months ago Was this helpful? --- # Email Protection | Cloudflare Please enable cookies. Email Protection ================ You are unable to access this email address doc.flowith.io ---------------------------------------------------------- The website from which you got to this page is protected by Cloudflare. Email addresses on that page have been hidden in order to keep them from being accessed by malicious bots. **You must enable Javascript in your browser in order to decode the e-mail address**. If you have a website and are interested in protecting it in a similar way, you can [sign up for Cloudflare](https://www.cloudflare.com/sign-up?utm_source=email_protection) . * [How does Cloudflare protect email addresses on website from spammers?](https://developers.cloudflare.com/waf/tools/scrape-shield/email-address-obfuscation/) * [Can I sign up for Cloudflare?](https://developers.cloudflare.com/fundamentals/setup/account/create-account/) Cloudflare Ray ID: **97144a299fef078c** • Your IP: Click to reveal 54.237.218.47 • Performance & security by [Cloudflare](https://www.cloudflare.com/5xx-error-landing) --- # Neo について | Flowith Official Document ![Page cover](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fzu8C1nUY5LDCaMxYYNnm%2F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png&width=1248&dpr=4&quality=100&sign=4dfcd71&sv=2) [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/neo-nitsuite#what-is-agent-neo) **What is Agent Neo?** ---------------------------------------------------------------------------------------------------------------------------- Agent Neo is a cutting-edge intelligent agent designed by the Flowith team to autonomously handle complex, multi-step tasks. It represents an evolution beyond static, pre-planned execution agents, operating with a dynamic and adaptive methodology. Key characteristics of Neo include: * **Autonomous:** Interprets high-level goals and works independently toward a solution. * **Dynamic:** Adapts its approach in real-time based on new information and the results of previous actions. * **Tool-Oriented:** Utilizes a diverse and expanding set of tools to gather information, process data, and execute actions. * **Self-Correcting:** Identifies failures or insufficient results and modifies its strategy to overcome obstacles and achieve the goal. Neo's external capabilities are continuously updated and upgraded, allowing it to become increasingly powerful over time. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2FLb1TccN0scGPrEBlzdlD%2FAmazing%2520Mockups.jpeg&width=768&dpr=4&quality=100&sign=50a313dc&sv=2) Neo Recipe and Nodes In other words, Neo is a **data-level** (operating at the data layer, without physical medium or real-world interaction) **generalized** (capable of performing general tasks, not limited to pre-trained or pre-defined functions) **Agent** (beyond just an LLM, able to utilize tools, manage memory, and collaborate with multiple LLMs) that can complete complex tas**ks** (beyond simple queries or commands) and adapt to new situations (capable of self-learning and processing in novel environments or tasks). [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/neo-nitsuite#core-philosophy-from-static-plans-to-dynamic-recipes) **Core Philosophy: From Static Plans to Dynamic Recipes** -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- The term "Recipe" originally refers to a set of instructions for preparing a dish. In the context of Neo, a Recipe represents a systematic workflow that AI executes to complete user tasks. When a user inputs a requirement or instruction, Neo breaks down the overall directive into smaller subtasks based on its capabilities and available tools. These subtasks are then passed to a task allocator for step-by-step execution. > A **Recipe** is a living, evolving set of instructions that Neo builds and refines as it works, not just a fixed list created at the beginning. This dynamic approach is more effective for complex problems where the full path to a solution is not knowable in advance, allowing Neo to navigate uncertainty and adapt to new discoveries. ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fzu8C1nUY5LDCaMxYYNnm%2F%25E4%25BA%25A712%25E5%2593%2581%25E5%259B%25BE%25205.png&width=768&dpr=4&quality=100&sign=4dfcd71&sv=2) A Recipe Generated by Neo [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/neo-nitsuite#the-anatomy-of-a-neo-task-core-concept) **The Anatomy of a Neo Task: Core Concept** ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- ### [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/neo-nitsuite#the-dynamic-recipe-neos-evolving-strategy) **The Dynamic Recipe: Neo's Evolving Strategy** A Recipe is the real-time sequence of steps Neo executes. It is generated step-by-step during the task, with each new action informed by the outcome of the previous one. Users can monitor this process through a live execution log, which provides transparency into Neo's "thought process" as it happens. The flow is cyclical and adaptive: Goal -> First Step -> Execution -> Result Analysis -> Next Step -> ... -> Completion ### [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/neo-nitsuite#the-dispatcher-and-tools-neos-capabilities) **The Dispatcher & Tools: Neo's Capabilities** The Dispatcher, also known as a Task Allocator, is a crucial component in the execution of subtasks generated by Neo. For each Neo-generated subtask, a dedicated Agent within the Dispatcher is responsible for allocation and implementation. This Agent identifies the most suitable tool from Neo's supported toolset and then executes and completes the subtask. For example, when the subtask is "**search for user reviews of flowith on social media**", the task allocator will invoke the "**online\_search**" and "**read\_single\_webpage**" to conduct searches for Flowith user reviews. You can view the currently supported tools for Neo on the Neo conversation initiation page. This list is updated in real-time, and we continuously add the most useful tools to Oracle's supported toolkit. [Oracle がサポートする現在の機能](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/oracle-gasaptosuruno) ![](https://doc.flowith.io/~gitbook/image?url=https%3A%2F%2Fcontent.gitbook.com%2Fcontent%2F2agAjJERs7Y6IiotYFD6%2Fblobs%2Fg6n2Kda04BzP3TPTHtPJ%2F%25E4%25BA%25A7%25E5%2593%2581%25E5%259B%25BE4.jpg&width=768&dpr=4&quality=100&sign=ebd383ce&sv=2) Tools supported by Neo ### [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/neo-nitsuite#intelligent-context-and-memory-management) **Intelligent Context & Memory Management** Complex tasks often generate extensive context. To maintain efficiency and accuracy, Neo employs an intelligent memory manager. For each step, the manager evaluates which information from previous steps is essential and injects only the necessary context. This prevents context overload and ensures the Large Language Model (LLM) can focus on the most relevant data for the current subtask. ### [](https://doc.flowith.io/flowith-heyoukoso/oracle-mdo-agent-mdo/neo-nitsuite#adaptation-and-self-correction) **Adaptation and Self-Correction** Neo's ability to self-correct is central to its power. The entire Recipe is fluid and can be altered mid-execution based on real-time results. This adaptation is triggered automatically by events such as: * **Tool Failure:** If a tool returns an error or no results, Neo will automatically try a different tool or rephrase its query. * **Unsatisfactory Results:** If the information gathered is insufficient, Neo will autonomously add new steps to find more data. * **New Discoveries:** Information found in one step can completely change the subsequent steps in the Recipe. > **Example of Adaptation:** > > 1. **Goal:** "Summarize recent AI hardware announcements." > > 2. **Step 1:** Neo uses > > Google Search > > for "AI hardware announcements 2025". > > 3. **Result:** Results are mixed, including news articles and one press release for a "Project Chimera" chip. > > 4. **Adaptation:** Neo identifies "Project Chimera" as a key term and creates a new sub-task to investigate it. > > 5. **Step 2:** Neo performs a new, targeted search for > > "Project Chimera" chip specifications > Last updated 1 month ago Was this helpful? ---