Find Where Latency Is Going with Noz

SigNoz released a new AI-powered feature called Noz that helps developers quickly identify the root cause of latency spikes by analyzing traces and decomposing them span by span. The tool allows users to ask natural language questions about endpoint slowdowns and automatically surfaces the bottleneck span or downstream dependency with suggested actions.

One endpoint got slow this afternoon and the rest of the service looks fine. Instead of opening the Traces view and sorting by duration yourself, you ask Noz to find where the time goes. Prerequisites - A SigNoz Cloud https://signoz.io/teams/ account with Noz https://signoz.io/docs/ai/noz/ available. - A service sending traces to SigNoz. Step 1: Ask where the time goes Open Noz from the top-right header and describe the slowdown: The /checkout endpoint got slow this afternoon. Where is the time going? Noz checks the endpoint's latency over the window, reports the p99 against its baseline, and tells you whether the time is spent in your own code or waiting on a downstream call. Step 2: Break it down to the bottleneck span Use Add Context → Services to attach the service, then go to the span level: Find the slowest trace for this endpoint and tell me which span or downstream dependency owns most of the time. Noz pulls a representative slow trace, decomposes it span by span, and attributes the latency to a specific operation or dependency, with Suggested Actions to keep going, such as pulling more traces for that operation or checking the dependency's own latency. Tips Ask for the tail, not the average. Request p99 or p95 so a few slow requests don't hide behind a healthy median. Follow the dependency. If the time is in a downstream call, ask Noz to look at that service next. Under the Hood Under the Hood under-the-hood To answer, Noz works through several agentic steps, visible under Worked through N steps : | Step | What It Did | |---|---| | Ran builder query | Ranked the service's operations by p99 latency over the window | | Searched traces | Pulled the slowest traces for the endpoint | | Reasoned | Decomposed a trace by span and attributed the time to a span or dependency | Next Steps Investigate What Changed After a Deploy with Noz https://signoz.io/docs/ai/use-cases/noz-incident-triage/ - When the slowdown started after a release. Latency Spike Explainer https://signoz.io/docs/ai/use-cases/latency-spike-explainer/ - Run the same breakdown from your coding agent. If you need help with the steps in this topic, please reach out to us on SigNoz Community Slack https://signoz.io/slack/ . If you are a SigNoz Cloud user, please use in product chat support located at the bottom right corner of your SigNoz instance or contact us at cloud-support@signoz.io mailto:cloud-support@signoz.io .