> For the complete documentation index, see [llms.txt](https://docs.maiagent.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.maiagent.ai/maiagent-user-guide/maiagent-user-guide-en/maigpt/deep-research.md). # Deep Research Deep Research is MaiGPT's automated research feature. Give it a topic, and AI will automatically search large volumes of web data, analyze and organize it, and produce a comprehensive research report. It's ideal when you need to deeply understand a topic but don't want to spend hours searching and organizing information yourself. *** ## How Is It Different from Regular Chat | | Regular Chat | Deep Research | | ---------------- | ---------------- | ----------------------------------------------------------- | | Speed | Seconds | Minutes | | Search Scope | Single search | Automatically searches 15-25 sources | | Response Length | A few paragraphs | Full report (4,000+ words) | | Source Citations | Sometimes | Always, with citations for every section | | Best For | Quick Q\&A | In-depth research, market analysis, technical investigation | *** ## When to Use It | Scenario | Example | | ------------------------ | ----------------------------------------------------------------------------------- | | Market research | "Research the current state and trends of the AI customer service market in Taiwan" | | Competitive analysis | "Analyze the product differences between Company X and Company Y" | | Technical investigation | "Research the latest developments and best practices in RAG technology" | | Policies and regulations | "Compile the latest progress on AI-related regulations in Taiwan" | | Industry reports | "Research global SaaS industry trends for 2025" | *** ## Starting a Research Session In just three steps, AI will automatically produce a complete research report. ### 1. Click the Deep Research Button Find and click the Deep Research button in the chat interface. ### 2. Enter Your Research Topic Describe your research topic in one or two sentences. The more specific, the better the report quality. | General | Better | | --------------------- | ---------------------------------------------------------------------------------------------------------------- | | "Research AI" | "Research the current state, challenges, and success stories of generative AI adoption in Taiwanese enterprises" | | "Analyze competitors" | "Compare MaiAgent, Dify, and Coze on their enterprise RAG capabilities" | | "Look up regulations" | "Compile 2025 AI-related regulations and regulatory guidelines in Taiwan" | ### 3. Wait for AI to Finish AI will begin researching automatically. You can see: 1. **Planning phase** — AI breaks the topic into sub-questions 2. **Search phase** — Searches the web for each sub-question 3. **Analysis phase** — Organizes and cross-references data 4. **Report phase** — Produces the full report and saves it to Canvas automatically The entire process typically takes 2-5 minutes, depending on topic complexity. You can work on other things — you'll be notified when it's done. {% hint style="success" %} **Give AI More Context** Beyond the topic, you can also tell AI: * Your role: "I'm a marketing director" * Report purpose: "This is for a weekly meeting presentation" * Special focus: "Focus on the Taiwan market" * Format requirements: "Use bullet points, don't exceed 3,000 words" {% endhint %} *** ## Reading and Exporting Reports After Deep Research completes, the report appears automatically in Canvas. You can read, edit, or export it directly. ### Reading the Report The report includes the following structure: | Section | Content | | -------------------- | -------------------------------------------------------------------------- | | **Summary** | Key findings and conclusions | | **Body** | Detailed analysis organized by topic in chapters | | **Source Citations** | Each section cites its data sources, with clickable links to original text | ### Modifying the Report After the report is generated, you can continue refining it in Canvas: | You Say | AI Does | | -------------------------------------------------- | -------------------------------- | | "The summary is too long, condense it to 3 points" | Shortens the summary | | "Add data charts" | Adds charts | | "Translate this section to English" | Translates the specified section | | "Reformat as a presentation outline" | Reformats the layout | ### Exporting Click the export button in the top right corner of Canvas: * **Copy** — Paste into Email, Word, Notion, etc. * **Download** — Save as a file {% hint style="success" %} **Paste Directly into Presentations** After copying, you can paste directly into PowerPoint or Google Slides with formatting preserved. This is perfect for quickly turning research results into weekly meeting reports. {% endhint %} *** ## Next Steps * [Continue editing with Canvas](/maiagent-user-guide/maiagent-user-guide-en/maigpt/canvas.md) * [Back to MaiGPT Overview](/maiagent-user-guide/maiagent-user-guide-en/maigpt/maigpt.md) --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.maiagent.ai/maiagent-user-guide/maiagent-user-guide-en/maigpt/deep-research.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.