This is an open-sourced, research-driven initiative designed to explore and develop multi-agent AI systems over the winter. The project focuses on building a General Purpose AI Agent Framework and establishing a Registry of Domain-Specific AI Agents.
- Intros
- AI Agent Project Overview
- Agent Presentation
- Agent Registry
- Agent Product Specs
- Codebase Setup
- Communications / Next Steps
- Q&A
- Framework: General Purpose AI Agent Framework
- Registry: A collection of specialized AI agents tailored to specific domains.
- Comprehensive guides for understanding and using the framework and registry.
- A user-friendly interface for showcasing the project.
- Frontend/UI: Intuitive design for interaction.
- Backend: Scalable and efficient architecture.
- Software Developer Kit (SDK): Tools for developers to build and integrate agents.
- A structured catalog of reusable, domain-specific AI agents.
- Contribute to the open-source agent repository or develop a closed-source agent to add to the registry.
- Deliver a Final Presentation during “AI Day,” showcasing:
- Video tutorials of AI agents in action.
- Research insights from the project.
The project enables various ways to deploy and use AI agent systems:
- Access through the product UI.
- Deploy via a desktop application.
- Use the developer kit to integrate into applications.
- NLP queries are routed to one or more agents (Agent 1, Agent 2, Agent 3, etc.).
- Tasks are performed either in series or in parallel.
- The system generates and returns an output to the user.
- Includes logs to monitor agent progress.
- Spawn agent templates directly into the chatbot.
- Use a Drag & Drop Flow to design and execute compound agent systems.
We will aim to really build two products over the winter:
- the general purpose AI agent framework (main project)
- a registry of specialized agents for specific domains
For each of these products, there will be a lot of details. For folks on the technical side, if you could try to follow the codebase development environment document in the materials folder of the Github repository, and get the codebase set up that would be appreciated. If you have any questions / experience any bugs, feel free to reach out directly, and I can either help you resolve the error or point you to the right POC.
For folks on the non-technical side, if you could look into the following resources that would be appreciated:
- Phidata - https://www.phidata.com/
- Crew AI - https://www.crewai.com/
- Magentic One - https://www.microsoft.com/en-us/research/articles/magentic-one-a-generalist-multi-agent-system-for-solving-complex-tasks/
- AutoGPT - https://github.com/Significant-Gravitas/AutoGPT
- Langgraph - https://www.langchain.com/langgraph
As far as a final deliverable, we will have final presentations on February 11th at 6 PM EST (date / time tentative). It would be awesome if whoever is interested could give a presentation on a topic that they worked on over the winter related to agents (similar to AI Day), that we could share / post on our youtube channel.
Natan will aim to hold weekly office hours on each Monday between 5 PM EST and 6 PM EST, in case anyone has any questions and would like to collaborate.
Communications will take place in the #agents thread within the anote slack channel (temporary link to join here): https://join.slack.com/t/anote-ai/shared_invite/zt-2vdh1p5xt-KWvtBZEprhrCzU6wrRPwNA
Join the initiative by contributing to the open-source repository, building specialized agents, or collaborating on research and development. Showcase your work during the AI Day event!
Note: to run code correctly you'll need to replace the openAI api key