Building an Scientific Research Team with Google's Agent Development Kit

This repository contains an example multi-agent system built using Google's Agent Development Kit (ADK) and powered by Google's Gemini models.

Scope

The goal of this project is to demonstrate how a team of AI agents can automate parts of the scientific literature review process. Specifically, this system aims to:

1. Search for relevant papers on bioRxiv based on a user's research topic.
2. Download selected papers.
3. Read the PDF content.
4. Extract specific information, such as research methodologies.
5. Summarize the findings for the user.

Architecture

This system employs a multi-agent architecture orchestrated by ADK:

• Root Agent (research_agent): Acts as the supervisor, coordinating the overall workflow.
• Search Agent (search_agent): Specializes in searching articles in bioRxiv.
• Methodologies Agent (methodologies_agent): Downloads papers, reads PDFs, and uses its LLM reasoning to extract methodologies from the text.

Running the Agent

1. Clone the Repository:

   git clone https://github.com/ericabelson/agentic-research-assistant.git
   cd agentic-research-assistant

2. Install Dependencies:

   pip install -r requirements.txt

3. Set API Key:

   • Add your Google AI Studio API key to the .env file:

     GOOGLE_API_KEY=YOUR_GEMINI_API_KEY_HERE
     

4. Run with ADK:

   • Web UI:

     adk web

     Navigate to http://localhost:8000 in your browser and select research-agent.
   • Command Line:

     adk run research-agent

     This starts an interactive chat session in your terminal.

Blog Post

Learn more about the design and implementation in our blog post: [Link to Blog Post]

Contributing

This is a sample project built with Agent Development Kit. To join the ADK open source movement, visit the Google ADK Python repository to get started, talk with other agent builders, or make your contribution!