Fedarch
Building onchain rules based agent orchestration.
Demo
About
The problem it solves It showcases the distributed protocol that lets efficient orchestration of the Multi agent protocols onchain with : 1. full persistence of the agent memory , the tasks and the tool calling in PII preserving fashion. 2. fast resolution of the agent authentification and creation of the various workflows without using low-code / other clumsy and less secure AI orchestration frameworks Use of AI tools and agents We used the Vercel AI react library in order to integrate our workflow with the frontend . also developed the custom AI provider to support both the orchestration of the local as well as proprietary AI agents.
How it's made
Challenges we ran into Challenge was to impleemnt the Multi agent orchestration framework with the persistence both onchain as well as off-chain of this protocol. Also another complexity is to fetch the history of interactions from the other AnythingLLM instances.