search

SciPhi-AI / R2R

Containerized, state of the art Retrieval-Augmented Generation (RAG) system with a RESTful API
https://r2r-docs.sciphi.ai/
MIT License
3.71k stars 280 forks source link

Multi-Agent Collaboration with R2R Framework #1632

Open regenrek opened 2 days ago

regenrek commented 2 days ago

Hello,

Does R2R support multi-agent collaboration, or any recommendatuon wit integrate an external framework? Im evaluation atm to buold a system where agents collaborate while leveraging R2R’s ingestion and retrieval capabilities, including GraphRAG.

Could you share recommendations or best practices for enabling agent communication and sharing context using R2R? Any examples or resources would be greatly appreciated.

Thank you!

NolanTrem commented 20 hours ago

Hey @regenrek, we haven't explored this extensively as a team but it would be really interesting to see R2R being used within a framework like crewAI. I think that the recommendation that I would have right now would be to give agents tools to interact with the R2R API.

Right now the agent that we have built in does something similar, by having a tool to call the RAG endpoint, but this is a single agent. You might look at some of the implementation there for inspiration: https://github.com/SciPhi-AI/R2R/tree/main/py/core/base/agent