Skip to Main Content

Agentic learning with a frozen agent is difficult. Agents can't reliably extract signal (cause of errors), and world-drift (the outcomes change due to the world of input changing) make it difficult to rely on old memories. For that I present my custom agentic harness-learner. It works by accumulating errors from agents that finish a job (in this case financial analysis), categorizing them and clustering them to learn shape, and then, after some accumulation uses an external oracle (in this case a strong agent that knows facts of the world) to create the rules. The memory corpus then has every agent use an adaptive hook-latch mechanism (think attention) that allows relevant rules to input case be injected into the model. Each rule features a full lifecycle, so rules can change and be forgotten

Team

Slavazin

More from The Harness Engineering & Model Wrangling #SFTechWeek Hackathon

MongoMation is an autonomous PLC engineer. You describe a machine in plain English, and Claude agents write IEC 61131-3 PLC code, compile and deploy it to a…

MongoMation project preview

LabRat is a self-improving harness for a physical lab agent, running on a real bench: two SO-101 arms, a 3018 CNC carrying a servo pipette I designed during…

LabRat project preview

Recursive Harness Scientist: an AI scientist that improves its own scientific method, and must replicate every gain on data it has never seen. Why it…

Recursive Harness Scientist project preview

Super Bid is a self-improving AI estimator that teaches itself to price bids the way a specific company would. It practices on the company’s real past bids…

SuperBid project preview

A robot that gets better at handling objects it’s never seen, without retraining. After every round, an AI coach reviews what broke and rewrites the robot’s…

WORLDSTATE gives physical spaces a memory. Any phone becomes an observer: open a link, point the camera, and WORLDSTATE tracks what's inside fridges, shelves…

WorldSTATE project preview