Google announced AQuA, an Ambient Quality Agent that watches AI agents in production. It reads recent session traces from Cloud Trace, Logging or BigQuery, grades them against a nine‑point checklist, clusters similar failures and verifies the top issues. The tool runs beside the agent, never in the request path, and can be triggered on demand or after each deployment.
Why it matters
It lets developers spot silent production bugs, like missed seat‑availability checks, without manually reviewing every conversation.
