Deeper Seeker

Overview

The Deeper Seeker is an open-source research tool similar to perplexity pro , designed to perform tasks like comprehensive market research, competitor analysis, and investment memo preparation. Building on the foundation of the previous version, Deeper Seeker V2.0.0 introduces significant improvements in research logic, web search capabilities, and report generation. The tool is designed to:

• Conduct iterative research with continuous refinement.
• Generate structured search queries and analyze results.
• Produce well-formatted, actionable reports with inline citations.

---

graph TD;
    %% Start and Initial Steps
    A[Start] -->|User Input| B[Run Follow-Up Loop]
    B -->|Generate Follow-Up Questions| C[Follow-Up Interaction]
    C -->|User Responds| D[Update Context]
    D -->|Check Iterations| D1{More Iterations?}

    D1 -- Yes --> B
    D1 -- No --> E[Generate Research Plan]

    %% Research Plan and Multi-Step Execution
    E -->|Break into Steps| F[Research Plan Steps]
    
    subgraph Research Execution
        direction TB;
        F -->|Step 1: Topic Exploration| S1[Generate Queries for Step 1]
        S1 -- Query 1 --> Q11[Fetch Results for Query 1]
        S1 -- Query 2 --> Q12[Fetch Results for Query 2]
        S1 -- Query 3 --> Q13[Fetch Results for Query 3]

        F -->|Step 2: Competitive Analysis| S2[Generate Queries for Step 2]
        S2 -- Query 1 --> Q21[Fetch Results for Query 1]
        S2 -- Query 2 --> Q22[Fetch Results for Query 2]
        S2 -- Query 3 --> Q23[Fetch Results for Query 3]

        F -->|Step 3: Market Trends| S3[Generate Queries for Step 3]
        S3 -- Query 1 --> Q31[Fetch Results for Query 1]
        S3 -- Query 2 --> Q32[Fetch Results for Query 2]
        S3 -- Query 3 --> Q33[Fetch Results for Query 3]
    end

    %% Processing and Report Generation
    Q11 --> P[Process Search Results]
    Q12 --> P
    Q13 --> P
    Q21 --> P
    Q22 --> P
    Q23 --> P
    Q31 --> P
    Q32 --> P
    Q33 --> P

    P -->|Aggregate Learnings| J[Extract Insights & Learnings]
    J -->|Refine Content| K[Generate Final Report]
    K -->|Format in Markdown| L[Save Report as final_report.md]
    
    L --> Z[End]

    %% Styling
    style A fill:#f9f,stroke:#333,stroke-width:2px;
    style Z fill:#f9f,stroke:#333,stroke-width:2px;
    classDef parallel fill:#ff9,stroke:#333,stroke-width:2px;
    class S1,S2,S3,Q11,Q12,Q13,Q21,Q22,Q23,Q31,Q32,Q33 parallel;

Loading

Key Features

1. Iterative Research Workflow:

   • Plans research steps based on user queries.
   • Generates precise search queries using the Exa API.
   • Continuously refines research based on findings.
   • Parallel execution of steps and search queries.

2. Structured Output:

   • Produces JSON-structured search queries for API calls.
   • Formats search results with highlights, citations, and summaries.

3. Comprehensive Reporting:

   • Synthesizes research findings into actionable reports using Google Gemini.

4. Multi-Model Support:

   • Utilizes multiple AI models (e.g., OpenAI, Groq, Google Gemini) for reasoning, planning, and report generation.

---

How It Works

1. User Query:

   • The user provides a research query (e.g., "Analyze the global EV market in 2024").

2. Follow-Up Questions:

   • The tool generates follow-up questions to refine the research scope and context.

3. Research Planning:

   • The AI creates a detailed research plan, breaking the query into logical steps.Each step leads to multi-query search.

4. Search Execution:

   • The tool generates specific search queries for each step and uses the Exa API to search the web for relevant information.
   • Parallel execution of search queries to reduce the latency.

5. Result Processing:

   • Search results are processed, formatted, and analyzed.

6. Final Report:

   • All findings are synthesized into a comprehensive, well-formatted report using Google Gemini.

---

Installation

1. Clone the repository:

   git clone https://github.com/HarshJ23/Deeper-Seeker.git
   cd Deeper-Seeker

2. Install dependencies:

   pip install -r requirements.txt

3. Set up environment variables:

   • Create a .env file and add your API keys:

     EXA_API_KEY=your_exa_api_key
     OPENAI_API_KEY=your_openai_api_key
     GROQ_API_KEY=your_groq_api_key
     GEMINI_API_KEY=your_gemini_api_key
     

---

Usage

1. Run the script:

   python app.py

2. Enter your research query when prompted:

   Enter your query: Analyze the competitive landscape of the cloud computing industry.
   

3. View the research process and final report:

   • The tool will display reasoning, plans, search results, and link counts for each iteration.
   • The final report will be saved as final_report.md.

---

Example Queries

Here are some sample queries to test the tool:

1. Market Research:

   • "Provide an overview of the global electric vehicle (EV) market in 2024."
   • "What are the current trends in the plant-based food industry?"

2. Competitor Analysis:

   • "Compare Tesla and Rivian in terms of market share and product offerings."
   • "Analyze the competitive landscape of the cloud computing industry."

3. Investment Memo Prep:

   • "Prepare a brief investment memo for a fintech startup specializing in blockchain-based payments."
   • "Evaluate the investment potential of the renewable energy sector."

---

Code Structure

• app.py: Main script for running the research assistant.
• generate_followup(): Function to generate follow-up questions based on user input.
• generate_research_plan(): Function to create a structured research plan.
• generate_queries_for_step(): Function to generate search queries for each research step.
• execute_plan(): Function to execute the research plan and fetch search results.
• generate_report(): Function to generate a detailed report using Google Gemini.

---

Dependencies

• Python 3.11+
• Libraries: refer requirements.txt file

---

Configuration

• Exa API Key: Required for web search functionality. Sign up at Exa AI.
• OpenAI API Key: Required for AI reasoning and planning. Sign up at OpenAI.
• Groq API Key: Required for using Groq's AI models. Sign up at Groq.
• Google Gemini API Key: Required for report generation. Sign up at Google Cloud.

---

Contributing

Contributions are welcome! Please follow these steps:

1. Fork the repository.
2. Create a new branch for your feature or bugfix.
3. Submit a pull request with a detailed description of your changes.

---

Future Enhancements

• In progress

  • Enhancing the overall workflow (reasoning,websearch and writing components).
  • Text pre-processing of extracted learnings to remove noise.

• Later Stage Additions:

  • FastAPI endpoint.
  • Support for multiple models.
  • frontend UI
  • Support for multiple model providers.
  • Database integration.

---

Happy researching! 🚀