# Meet Noz, your AI teammate inside SigNoz > Source: > Published: 2026-06-26 00:00:00+00:00 # Meet Noz, your AI teammate inside SigNoz Last month, we wrote about [bringing agent-native observability to SigNoz](https://signoz.io/blog/introducing-agent-native-observability/). The idea was simple. The platform we spent years building is open source, OpenTelemetry-native, and built around logs, metrics, and traces correlated in one database. That same foundation also makes SigNoz a strong place for AI agents to work with observability data. The hosted MCP server was the first step. Today we are shipping the next one. Meet Noz, your AI teammate inside SigNoz. Ask it about your telemetry or how SigNoz works, and it answers from your own data and the docs. Give it an alert or dashboard request, and it helps build it for you. Noz is in beta and free to use while it is in beta. Why we built Noz We did not start with *"we should add an AI chatbot."* We started with what users were already doing. SigNoz has had human chat support built into the product for a long time. Over the past year, users began treating that chat like an AI. They asked how to build dashboards, why ingestion looked off, and how to write queries for specific services. The demand for an in-product assistant was already there. They expected instant answers, the way you would from an AI, and some even apologized when they realized a person had been replying all along. That told us two things. Users want answers inside SigNoz, at the moment they are stuck, without leaving for the docs or a separate tool. And the questions they ask map to the same mechanical steps that slow everyone down, like finding the right data, remembering attribute names, and turning an intent into an alert or a dashboard. So we built Noz to remove that friction, not to bolt an AI chat onto the product because everyone else is doing it. Noz is a teammate, not a replacement for your engineers. It does the grunt work, finding the data, writing the query, building the artifact, while you stay in control of the decisions. When it handles the mechanical parts, you spend less time getting to signal. What you can do with Noz Noz lives in a sidepane next to whatever you are looking at, so it has context on the page you are on. The easiest way to see what it does is create an alert. Tell Noz "alert me when checkout p99 latency goes above 2 seconds" and it builds the alert rule. The same pattern applies across SigNoz. You can describe a dashboard and have Noz assemble the panels, or ask about error rates, slow traces, and log volume for a service, then jump into the explorer when you need to dig deeper. These are a few examples. Start with whatever you are trying to get done in SigNoz, and you might be surprised by how much it already handles. Noz, or the MCP server with Claude and Codex? This is a fair question, especially since we shipped the [hosted MCP server](https://signoz.io/docs/ai/signoz-mcp-server/) last month. Both connect AI to your SigNoz data, but they serve different moments, and we use both ourselves. Reach for Noz when you are working inside SigNoz and want zero setup. It is already there in the sidepane, authenticated with your session, aware of the page you are on, and able to deep-link results back into the explorer. That makes Noz a good fit for anything that runs through the SigNoz UI, which an agent working from your terminal cannot drive. Reach for the MCP server with Claude Code, Cursor, or Codex when your investigation reaches past SigNoz. If you need to correlate a spike with a recent commit, a GitHub issue, or a Slack thread, an external agent with the MCP server can pull SigNoz data alongside your other tools in one place. Both use the same open foundation. Noz is the faster path when the work starts inside SigNoz, while the MCP server connects SigNoz data to tools outside the product. What's next Noz is focused today on work inside SigNoz, and we are building in two directions next. **Bring more investigation context into SigNoz.** Important investigations often involve more than telemetry. As Noz connects to tools your team already uses, it can pull that context into the investigation instead of leaving you to stitch it together by hand. **Handle more incident-response work.** Noz already helps investigate by correlating signals and narrowing likely causes. We are improving that flow so it can handle more repetitive incident tasks and help reduce MTTR. We are also building toward alert-triggered investigations, so Noz can start gathering context when an alert fires instead of waiting for you to ask. Through all of this, Noz keeps the same role, handling more of the repetitive work while engineers make the calls. That is what [agent-native observability](https://signoz.io/blog/introducing-agent-native-observability/) looks like in the product. Get started Noz is in beta and live in your SigNoz Cloud workspace, in the sidepane and on a full-screen page of its own. There is nothing to install, so open Noz and ask your first question. It is free while it is in beta. When it leaves beta, pricing will be transparent and easy to reason about, the same way we approach the rest of SigNoz pricing. A good first thing to try is to ask Noz to create an alert for a service you care about. For a fuller walkthrough, see the [Noz docs](https://signoz.io/docs/ai/noz/). And if a teammate would benefit, invite them in so they can use it too. Noz is now in beta inside SigNoz Cloud. Ask it to create alerts, build dashboards, and get answers from your telemetry and docs. [Get Started - Free](https://signoz.io/teams/)