Autoresearch: The feedback loop behind self-improving agents
Source
latent.space
Author
Richard MacManus
Date
Why it matters
Names the layer above the agent harnessThe scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.Full definition → — the evalA repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.Full definition →-and-feedback loop that keeps a deployed AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → improving — and argues for owning it in Git rather than renting it from a model provider.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
agent harness — The scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.
articleInside-Out AI: Rebuilding Airbnb Behind the Scenes and Across the Guest Experience
articleRunway’s WorldPrompt and the Engineering of Real-Time Worlds
articleCan Skills Learned in Games Transfer to Real-World Work?
articlePRs NOT Welcome: How Top AI Open Source Projects Are Managing Thousands of Contributors
