Noz: SigNoz's AI Teammate for Observability

Noz is SigNoz's AI teammate, built into the SigNoz UI. Ask about your traces, logs, metrics, dashboards, alerts, and infrastructure in plain English, and Noz investigates across your telemetry and replies with answers grounded in your real data. It works through a problem step by step, points to the likely root cause, and suggests a next action, without you writing queries or leaving SigNoz. Noz can also take that next step for you, like creating a dashboard, setting up an alert, or saving a view.

Open Noz

Click Noz in the top-right of the SigNoz header from any page. Noz opens as a side panel next to your current view, so you can keep your dashboard, trace, or logs list in context while you ask.

The panel header gives you a few controls:

• Recent conversations: jump back into an earlier conversation.
• New conversation: start a fresh thread.
• Expand: open Noz full screen, with a searchable conversation list on the left and more room to read long answers.
• Close: dismiss the panel.

Ask your first question

A new conversation opens with starter prompts so you don't have to think of one. Click any of them to run it instantly:

• Show me the top errors in the last hour
• What services have the highest latency?
• Give me an overview of system health
• Find slow database queries
• Which endpoints have the most 5xx errors?

Or type your own question in the Ask anything box (use Shift+Enter for a new line) or use voice input, then send it. Noz investigates and responds.

Add context

Use Add Context below the input to point Noz at a specific entity before you ask. You can attach:

• Dashboards: ask Noz to explain or diagnose a specific dashboard.
• Alerts: ask why an alert is firing, or whether it is noisy.
• Services: focus the investigation on a single service.

Adding context narrows Noz's scope so its answer covers only the dashboard, alert, or service you care about.

Understand and act on an answer

Noz doesn't just give you a conclusion. It shows the steps it took to get there and usually offers one-click follow-ups, so you can trust what you're seeing and keep moving without breaking flow. Expand the section below to see how Noz works through a problem and what it lets you do next.

How Noz investigates

Noz works as an agent: it runs multiple steps against your data, then reasons over the results. On answers that required investigation, expand Worked through N steps to see the trace. For example: think → decide what to query → run builder queries across traces, logs, and metrics → pull representative traces or logs → reason to a conclusion.

Because every step runs against your live telemetry, Noz tailors each answer to your environment.

Example: investigating what changed after a deploy

You: Errors on the checkout service jumped after the latest deploy. What changed?

Noz traces a failing request end to end, identifies which dependency is returning errors, and checks whether the failures line up with the new release, then reports back something like:

Checkout's POST /checkout is returning 502s, and they're sustained across the window, all on v2.0.0. The trigger is downstream, not checkout itself:

• The failure is in one dependency. The outbound external_api_call to external-svc gets an HTTP 500 from https://httpbin.org/status/500; POST /checkout then surfaces it to the caller as a 502.
• Checkout isn't broken. It's faithfully propagating a downstream 500 as a 502.
• The errors are concentrated in external-svc. The other three dependencies in the checkout path are clean.

Worth confirming whether the v2.0.0 release pointed checkout at that external-svc endpoint (or changed how its 500s are handled), since that's the only failing leg.

It then points to the likely cause and offers Suggested Actions (here, Check external-svc, Alert on 5xx, and Confirm regression) to keep investigating.

Suggested actions and follow-ups

After most answers, Noz shows Suggested Actions, one-click follow-ups tailored to the result (for example, Pull failing traces, Compare to the previous deploy, or Adjust the alert threshold). They save you from typing the obvious next question and keep the investigation moving. When a suggested fix would change your configuration, Noz asks for confirmation first ("Want me to do that?"), so you stay in control of any change.

Manage your conversations

Every conversation is saved, so an investigation you started this morning is still there when you come back to it.

• Pick up where you left off. Open Recent conversations in the panel header to reopen a thread, or hit Expand for the full-screen view, where conversations are grouped by recency (Today, Last 30 days) with an Archived section.
• Find a past investigation. Use Search in the full-screen view to pull up an old conversation by what you asked.
• Keep your list focused. Start a new conversation for an unrelated question so each thread stays on one topic, and archive the ones you're done with.

On any answer you can copy it to drop into a ticket or share with your team, regenerate it for another pass, and use thumbs up / thumbs down to flag when an answer was helpful or off. Your feedback trains Noz to give better answers over time.

Noz vs. the SigNoz MCP Server

Noz and the SigNoz MCP Server are two ways to bring AI to your observability data:

• Noz is the AI teammate inside the SigNoz UI. Reach for it when you are already in SigNoz and want to explore data, understand a dashboard, or triage an incident without writing queries. It can also build a dashboard, alert, or view for you on the spot.
• The SigNoz MCP Server connects external coding agents (Claude Code, Cursor, and others) to SigNoz, so you can query telemetry and create dashboards, alerts, and views from your IDE or terminal, right alongside your code.

Both are grounded in the same unified telemetry. Use whichever fits where you are working.

Next steps

If you need help with the steps in this topic, please reach out to us on SigNoz Community 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.