Agents and CodeMedium importance
Google open-sources AQuA, an agent that combs through other agents' production sessions to find silent failures and their cause in the code
Google released as open source AQuA (Ambient Quality Agent), a reference implementation that analyzes AI agent sessions in production, confirms recurring failure patterns and investigates the cause in the deployed code. The target is errors that ordinary monitoring misses, such as agents that ignore user preferences or skip mandatory tool calls and still return a successful response. It runs alongside the agent on Google Cloud, reading data from Cloud Trace, Cloud Logging or BigQuery, analyzes up to 1,000 recent sessions per run and, to find the cause, compares the failing sessions with a snapshot of the code saved at deployment time, citing file and lines and proposing changes. In a test with 32 sessions, developers cut from 15 to 2 the cases in which the agent skipped seat selection and from 7 to 0 those that forgot the dietary preference.
Published in edition no. 28 of Noticiário do Joãozinho ·
Translated by Joãozinho AI from the original Brazilian Portuguese edition. Facts, numbers and sources are the same as in the original; quotes, video titles and source names stay in their original language. Read the original in Portuguese
Original source
https://www.infoq.com/news/2026/11/google-ambient-agent-prod/