Automate pattern discovery with Topics, now generally available

Braintrust has released Topics, an active observability tool that automatically analyzes production AI traces to identify patterns and classify them across built-in dimensions like task, issues, and sentiment. The feature uses embedding-based clustering and LLM-assisted naming to create SQL-queryable labels that integrate into the broader evaluation and improvement workflow. Topics is now available on all pricing plans, enabling teams to move from discovering failure modes directly into testing and deploying fixes without manual intervention.