Deep Research Agent

A powerful AI-powered research assistant that conducts comprehensive research on any topic.

• Topic Breakdown: Automatically breaks down complex topics into relevant subtopics
• In-depth Research: Conducts detailed research on each subtopic
• Executive Summary: Generates a concise summary of all findings
• Adjustable Depth: Choose between light, medium, or deep research depth
• User-friendly Interface: Simple web interface to request and view research

1. The agent takes a research topic and desired depth level
2. It uses GPT-4 to break down the topic into subtopics
3. Each subtopic is researched in detail
4. An executive summary is generated from all findings
5. The complete research report is returned

• deepResearchAgent.ts - The main research agent (HTTP endpoint)
• researchInterface.ts - Web interface for interacting with the agent
• testAgent.ts - Test file to verify the agent is working correctly

The easiest way to use the Deep Research Agent is through the web interface:

1. Visit the Research Interface
2. Enter your research topic
3. Select the desired research depth
4. Click "Start Research"
5. View the comprehensive research results

You can also call the research agent directly via its API endpoint:

POST https://deepResearchAgent.YOUR_USERNAME.val.town

Request body:

{
  "topic": "Your research topic",
  "depth": "medium" // Options: "light", "medium", "deep"
}

Response:

{
  "success": true,
  "topic": "Your research topic",
  "research": {
    "topic": "Your research topic",
    "executiveSummary": "...",
    "subtopics": [
      {
        "title": "Subtopic 1",
        "content": "...",
        "wordCount": 500
      },
      // More subtopics...
    ],
    "generatedAt": "2023-06-15T12:34:56.789Z",
    "depth": "medium"
  }
}

To verify that the agent is working correctly, you can use the test endpoint:

GET https://testAgent.YOUR_USERNAME.val.town

This will run a simple test query and return the results.

If you encounter issues with the Deep Research Agent, try the following:

1. Check OpenAI API Access: Ensure your Val Town account has access to the OpenAI API.

2. Verify Endpoints: Make sure you're using the correct URLs for your Val Town username.

3. Check Console Logs: Open your browser's developer console to see any error messages.

4. Try a Simpler Topic: Start with a simple, well-known topic to test the agent.

5. Reduce Research Depth: Try using "light" depth which requires fewer API calls.

6. Check Response Format: If you're calling the API directly, ensure your request format is correct.

7. Run the Test Endpoint: Use the test endpoint to verify basic functionality.

You can modify the agent to suit your specific research needs:

• Adjust the number of subtopics based on depth in the generateSubtopics function
• Change the word count for each subtopic in the researchSubtopic function
• Modify the system prompts to focus on specific aspects of research

This agent uses the OpenAI API via Val Town's standard library. No additional setup is required.

• Research is based on AI knowledge, not real-time web searches
• Very complex or niche topics may have limited information
• Research depth affects processing time and token usage