Autonomously investigate a financial agent's failures end to end: pull traces, evals, and annotations; identify issues; propose and apply code fixes; build datasets; and run experiments to validate changes.

🔒  This use case is in the locked catalog. The description, method, result and source video — the free sample is the first 50; the rest come with a library pass. Unlock