torrresagus

Joined 20 days ago

Karma

Submissions

Member since

Oct 21

AI engineer in Buenos Aires. I build multi-agent systems: at Pomelo (fintech) I designed an "agent as a gateway" layer that routes each user to specialized agents, with memory, permissions and per-agent cost tracking. Before that I built enterprise agent platforms at EY, and I teach building LLM apps with LangChain at UTN.

Now building Callsign (https://callsign.im), an open-source network where AI agents from different companies message each other, with consent and an audit trail. Also co-founder of Brauni (https://brauni.io), software for psychologists.

https://www.linkedin.com/in/torres-agustin/

View on Hacker News

Recent submissions

Ten weeks after Groq shut down its models, GitHub still finds them in 314k files tiltmeter.dev

1 torrresagus 14 days ago 0 comments

Comment · 15 days ago

Nice launch. You mention agent delegation, so I'm curious how far that goes. The one-hop case is answered upthread (agent's grants ∩ user's). But what happens when my agent asks someone else's agent to do part of the job? Does that second agent act on my permissions, on my agent's, on its own, or on the overlap? And does the audit log show the whole chain, or just whoever made the last call?

Asking because I keep running into this with agents that talk to other people's agents, and the audit part worries me more than the permissions part.

torrresagus

Recent submissions

Ten weeks after Groq shut down its models, GitHub still finds them in 314k files t tiltmeter.dev

Ten weeks after Groq shut down its models, GitHub still finds them in 314k files tiltmeter.dev