Tracking: known patterns
Tracking holds pinned collections you monitor over time, the failure patterns you already know matter. Once a pattern is pinned, you can watch how often it recurs as new traces come in.Discovery: unknown patterns
Discovery is a chat interface over your pulled traces. Ask it questions about your data, let it surface patterns you didn’t know to look for, and pin the interesting ones into a Tracking collection.Traces: the raw table
Traces is a paginated table of every ingested trace (model, latency, tokens, preview). Bulk-select any subset and push it straight into a seed dataset for generation.Getting traces in
Findings connects to your observability tool (Langfuse or LangSmith today) and pulls real LLM traces in. See Connect AI Observability for setup, and Signals & Journeys if you also want user events correlated with those traces.Next steps
Reviews
Route the patterns you find to a human reviewer