Skip to main content

AG2

AG2 is a framework for building multi-agent applications with LLMs. This example uses AG2 to orchestrate the GPT Researcher multi-agent workflow.

Check out our blog posts:

Use case​

By using AG2, the research process can be significantly improved in depth and quality by leveraging multiple agents with specialized skills. Inspired by the recent STORM paper, this example showcases how a team of AI agents can work together to conduct research on a given topic, from planning to publication.

An average run generates a 5-6 page research report in multiple formats such as PDF, Docx and Markdown.

Please note: This example uses the OpenAI API only for optimized performance.

The Multi Agent Team​

The research team is made up of 8 agents:

  • Human - The human in the loop that oversees the process and provides feedback to the agents.
  • Chief Editor - Oversees the research process and manages the team.
  • Researcher (gpt-researcher) - A specialized autonomous agent that conducts in depth research on a given topic.
  • Editor - Responsible for planning the research outline and structure.
  • Reviewer - Validates the correctness of the research results given a set of criteria.
  • Revisor - Revises the research results based on the feedback from the reviewer.
  • Writer - Responsible for compiling and writing the final report.
  • Publisher - Responsible for publishing the final report in various formats.

How it works​

AG2 Pipeline

Stages:

  1. Planning stage
  2. Data collection and analysis
  3. Review and revision
  4. Writing and submission
  5. Publication

How to run​

  1. Install required packages: 
    pip install -r requirements.txt
    pip install -r multi_agents/ag2/requirements.txt
  2. Update env variables: 
    export OPENAI_API_KEY={Your OpenAI API Key here}
    export TAVILY_API_KEY={Your Tavily API Key here}
  3. Run the application: 
    python -m multi_agents.ag2.main

Usage​

To change the research query and customize the report, edit multi_agents/ag2/task.json.

Task.json contains the following fields:​

  • query - The research query or task.
  • model - The OpenAI LLM to use for the agents.
  • max_sections - The maximum number of sections in the report. Each section is a subtopic of the research query.
  • max_revisions - Maximum reviewer/reviser loops per section.
  • include_human_feedback - If true, the user can provide feedback to the agents. If false, the agents will work autonomously.
  • publish_formats - The formats to publish the report in. The reports will be written in the outputs directory.
  • source - The location from which to conduct the research. Options: web or local. For local, please add DOC_PATH env var.
  • follow_guidelines - If true, the research report will follow the guidelines below. It will take longer to complete. If false, the report will be generated faster but may not follow the guidelines.
  • guidelines - A list of guidelines that the report must follow.
  • verbose - If true, the application will print detailed logs to the console.