TraceTriage

TraceTriage is a web app (with optional CLI agent) that turns noisy distributed tracing and logs into a single, prioritized incident narrative. Instead of dumping raw spans, it automatically groups failures by “blast radius” (users impacted, endpoints, regions, deploy versions) and highlights the smallest set of services and recent changes most likely responsible. It integrates with OpenTelemetry collectors and common backends (Datadog, New Relic, Grafana Tempo/Loki) to avoid replacing your stack. The product focuses on the first 30 minutes of an incident: deduplicating alerts, correlating traces with deploys and feature flags, and generating a short, actionable checklist for responders. It’s an AI+traditional app: deterministic correlation and scoring plus an LLM to summarize evidence and propose next steps. Realistically, it wins only if it’s faster and cheaper than adding more dashboards and on-call headcount.