PyAI Research Assistant Agent

A powerful research assistant that generates detailed reports on any topic using PydanticAI and Tavily's search API. The assistant uses advanced AI to create well-structured, properly cited research reports with real-time progress updates.

Features

Search and Research

• Advanced semantic search using Tavily API
• Intelligent source validation and citation
• Real-time search result processing
• Automatic content summarization

Report Generation

• Well-structured research reports with proper citations
• Smart section organization based on available sources
• Automatic source validation and verification
• Hyperlinked citations and references

User Interface

• Modern Streamlit web interface
• Clean and intuitive design
• Real-time progress updates
• Raw markdown preview option
• One-click report downloads

Output Management

• Markdown-formatted reports
• Automatic report saving with timestamps
• Organized report storage in dedicated directory
• Rich console output with progress indicators

Error Handling

• Graceful error recovery
• Clear error messages and warnings
• Smart retry mechanism for failed requests
• Comprehensive error logging

Installation

Prerequisites

• Python 3.9+
• Git
• Conda (Recommended)

Setup

1. Clone the repository:

git clone https://github.com/dynamikapps/pyai-research-assistant-agent.git
cd pyai-research-assistant-agent

2. Create and activate a conda environment:

conda create -n research-assistant python=3.9
conda activate research-assistant

3. Install the required packages:

pip install -r requirements.txt

4. Create a .env file from the example:

cp example.env .env

5. Add your API keys to the .env file:

TAVILY_API_KEY=your_tavily_api_key
OPENROUTER_API_KEY=your_openrouter_api_key

Usage

Command Line Interface

Run the research assistant from the command line:

python research_assistant.py

Web Interface

Run the Streamlit web interface:

streamlit run streamlit_app.py

Output Structure

The research assistant generates well-structured reports in markdown format:

# Research Title

_Generated on YYYY-MM-DD HH:MM:SS_

## Introduction

[Introduction content with context and scope]
[Citations and hyperlinks to sources]

## [Body Section 1]

[Section content with detailed analysis]
[Citations and hyperlinks to sources]

## [Body Section 2]

[Section content with detailed analysis]
[Citations and hyperlinks to sources]

## Conclusion

[Summary and key findings]
[Citations and hyperlinks to sources]

## Sources

1. [Source Title](URL)

   > Relevant quote or summary from the source

2. [Source Title](URL)
   > Relevant quote or summary from the source

Report Features

• Clear section organization
• Proper citation formatting
• Hyperlinked references
• Source summaries and quotes
• Automatic timestamp
• Consistent markdown styling

Reports are automatically saved in the reports directory with timestamps in the format: research_report_YYYYMMDD_HHMMSS.md

API Keys

• Get your Tavily API key from: https://tavily.com
• Get your OpenRouter API key from: https://openrouter.ai

Contributing

Feel free to open issues or submit pull requests for improvements.

License

MIT License - see LICENSE file for details.