Grafana Assistant Investigations, Grafana Assistant Workspace, Grafana Assistant Automations, the Grafana Cloud MCP server, gcx, and Grafana Agent Observability all reach general availability during Grafana Labs' inaugural AI Week

Grafana Labs, the company behind the open observability cloud, today announced the general availability of six AI capabilities during its inaugural AI Week, extending Grafana Assistant into an agentic operations layer that detects, investigates, and remediates production issues at the pace AI now creates them. The releases include Grafana Assistant Investigations, Grafana Assistant Workspace, Grafana Assistant Automations, the Grafana Cloud MCP server, gcx, and Grafana Agent Observability.

Observability has traditionally started after code reaches production: instrument it, dashboard it, alert on it, and hope you notice issues before your customers do. That timeline is moving earlier. Engineers are shipping more code, faster. Agents have dramatically increased the rate of change in many teams, and the practices that keep systems reliable must move earlier in the lifecycle to keep pace. According to Grafana Labs' 2026 Observability Survey, 92% of practitioners say they'd get real value from AI catching anomalies, yet only 57% say they're currently implementing observability for their own AI systems in any capacity. Grafana Labs' answer to that gap is an assistant that brings high-quality insights from production telemetry right when you’re planning changes, not just after something breaks.

"We used to treat observability as something you bolt on just before code reaches production," said Mat Ryer, Senior Director of AI, Grafana Labs. "That's changing. Now, Grafana Assistant can review your plans before you've written a line of code, add the instrumentation for you once you have, watch features as they land in production, and stay with you as you scale while dealing with the inevitable incidents that follow. It's the same assistant end-to-end, and it doesn’t just advise; it acts.”

Planning before a single line of code ships

Grafana Assistant Workspace, now generally available, gives that first conversation a home: chat history, live canvas, and investigation reports in one place instead of a sidebar. Bring in an architecture before a service exists, and Grafana Assistant flags where it won't scale — before anyone opens an editor. Paired with Grafana Assistant Investigations, a running investigation becomes a shareable report, no rewriting required.

Turning a plan into instrumented, running code

Once a plan is in shape, Grafana Assistant helps engineers act on it without losing speed. Ask Grafana Assistant to instrument a new service, and it opens a pull request with the instrumentation in place, wires up the data source, configures Grafana, and watches until telemetry arrives, iterating with you if it doesn't. Two GA releases power this stage: gcx and the Grafana Cloud MCP server.

gcx, Grafana's agentic CLI for managing dashboards, alert rules, data sources, and other resources as code, brings self-managed Grafana and Grafana Cloud under one command line, with agent-friendly input and output for tools like Claude Code, GitHub Copilot, and Cursor, and full GitOps support for versioning dashboards and alerts in Git.

“Root cause analysis and dashboard creations are prompts in the Grafana Assistant side-panel that help us understand anomalies and correlations across signals that lead to speeding things up,” said Oren Lion, Director of Software Engineering, TeleTracking. “Now, with gcx, anyone can use Grafana Assistant in their coding agent, for example, to investigate flapping alerts, that's democratizing...Grafana!”

The Grafana Cloud MCP server gives any MCP-compatible client, including Claude Desktop, Cursor, and more, direct access to a Grafana instance's dashboards, alert rules, incidents, and data sources. No more pasting metrics into a chat window; the agent queries live telemetry from where the code is being written.

Grafana Assistant investigates while you stay in control

In production, the questions change. Instead of parsing raw metrics and logs, engineers can ask Grafana Assistant questions about their telemetry in plain language, learn what a signal is actually telling them, and plan their next move by correlating technical data with business outcomes.

Grafana Assistant Investigations and Grafana Assistant Automations are both now generally available, helping engineers get answers out of production faster than they can type their questions. When an incident happens, Grafana Assistant Investigations forms hypotheses about what went wrong and swarms over the problem to chase down the leads in the data, proving or disproving each one. Engineers can stay in the driver's seat and steer the investigation themselves, or step back and wait the few minutes it takes for Grafana Assistant to finish and hand back a conclusion. Grafana Assistant Automations let a saved Grafana Assistant prompt run again automatically, on a schedule, or on demand, so recurring checks like a daily error rate summary are sent to your Slack channel without anyone re-typing the same question every morning.

“When I’m in Grafana, it’s never a leisurely thing. It’s always under pressure; trying to find answers to unexpected issues in a very reactive situation,” said Rob Kacic, Engineering Manager, MasterControl. “Grafana Assistant Investigations has been super helpful with that. I’ve had cases where it cut my time drastically; correlating backend errors and traces to specific front-end page IDs would have taken me hours, but with Grafana Assistant Investigations, it was a 15-minute task. Lowering that bar to entry has been the biggest win so far.”

Observing the AI agents that you build

Grafana Agent Observability, generally available today, extends Grafana Cloud's OpenTelemetry-native monitoring to the AI systems teams are now shipping. It was born out of necessity: Grafana Labs needed to observe Grafana Assistant itself once usage started spiking past expectations. Instrumented agents emit telemetry that captures the usual signals (usage, latency, and errors) alongside two that matter specifically for AI systems: token usage/cost and the conversation itself. Conversation data gives teams forensic-level debugging down to the individual exchange. Built-in evaluators go a step further, letting teams introspect conversations to test a sample of conversations for the agent behavior directly, catching hallucinations, drift, or policy violations before a customer does.

“After two days in production [with Grafana Agent Observability], the difference was amazing,” said a spokesperson for Alter Domus. “We started collecting information about our MCP servers, prompts, and conversations, so now we can understand why MCP servers fail and why users are not achieving what they expect when interacting with those systems.”

“AI should allow you to move at 10x velocity, not produce incidents at 10x the rate,” Ryer said. “That's why we're releasing all these AI capabilities in a single week. Engineers shouldn't have to wait for their tools to catch up to how quickly they're already moving.”

Availability

• Assistant Investigations, Assistant Workspace, Assistant Automations, the Grafana Cloud MCP server, gcx, and Agent Observability are all generally available today.
• Try the tools with Grafana Cloud, including the free-forever tier.

About Grafana Labs

Grafana Labs, the company behind the open observability cloud, is founded on the principles of open source, open standards, open ecosystems, and open culture. Grafana Cloud, our fully managed observability platform, is flexible and built for scale. With Grafana Cloud's actually useful AI, organizations can see, understand, and act on all their disparate data to move at the speed of their ambitions, while getting the visibility they need to run AI systems reliably and at scale. Today, more than 35 million users and 7,000+ customers – including Anthropic, Bloomberg, NVIDIA, Microsoft, and Salesforce – trust Grafana Labs to ensure reliability of their applications and systems, resolve incidents quickly, and optimize their telemetry to reduce noise and cost. We are a 100% remote company with 1,600+ team members across 40+ countries, and we’re backed by leading investors including Lightspeed Venture Partners, Sequoia Capital, GIC, Coatue, J.P. Morgan, CapitalG, and Lead Edge Capital. Learn more at grafana.com and follow us on LinkedIn and X.

View source version on businesswire.com: https://www.businesswire.com/news/home/20260727216919/en/