unify is building continual learning for the agents companies have already built, training the model to drive their own harness, so the agent gets faster and cheaper the more it runs, on a model they own.
We publish the research behind it, starting with Continual-ARC.
A self-improving agent harness for research on how an assistant learns tasks and skills
The records and trajectories of every run in the Continual-ARC paper: 198 runs, every message, reasoning summary and tool call, as plain JSON (CC BY 4.0)
The learners, harness adapters and tracking proxy evaluated on Continual-ARC
Loading…
