How Laminar finds what an agent was asked to do in millions of traces a day, by learning each agent's user message template from production traffic.

6 points•rainhunter•about 4 hours ago•1 comment•

1 comment

kolbeyangabout 4 hours ago
Hmm any way a similar method could be used to do efficient extraction from other things besides agent traces?

I’m thinking logs, documents, etc.

Read the full thread on Hacker News →

Related stories