how does @faircaster works?

under the hood it works as team of agents, each with specialised roles and tools for the job, all coordinated by a orchestrator agent.

what can it do?

it tracks casts from individuals with the highest network centrality and quality (using @openrank). it uses these to find new projects as well as important new information about known projects.

and all that informs its investment strategy

Faircaster is a VC agent that finds Farcaster projects early

I get it now. So research agent checking the feed for new projects with the aid of openrank to sift through the noise. gg good sir

Barista in the morning, AI engineer by day, and parent (with @shiwen) by night ||
Host at /clickcast /cpg

yes, will have a lot more though. next module with core elements already built are tracking new accounts followed by top accounts (initially defined through influence, later through pnl)

Sick. And are you doing it multi agent or just have one generalist agent doing all these steps?

multi-agent, using lang graph for orchestration

modular is good to a) maintain and improve individual performance and b) add new modules over time

I like langgraph as well, use it mostly in my day job because I need the agents to follow deterministic flows, especially when dealing with sensitive processes like pricing etc.

makes sense. i'm team "roll your own framework" b/c all the frameworks have some sort of gotcha. in my research though langgraph was the least bad of them. at least you could point to at scale use cases of them like at klarna and linkedin