Grafana is attempting to close the AI observability gap before enterprise agents reign supreme

Observability startup Grafana Labs Inc. said today it’s attempting to shine a lightweight on the “black box” inner workings of artificial intelligence models with the launch of latest capabilities that can higher enable corporations to trust and control them in production.

The announcement got here during Grafana’s annual user conference GrafanaCON 2026 in Barcelona, where it also revealed it’s making a dedicated “AI organization” that will likely be led by its recent Director of AI Mat Ryer.

Grafana said vast numbers of enterprises are racing to integrate AI agents into their workflows with a view to enhancing automation, but doing so presents quite a few challenges by way of observability. Initially is the undeniable fact that AI applications behave very in a different way compared with the standard software that helped Grafana establish itself as a significant observability player.

Because they work so in a different way, existing monitoring tools struggle to supply much insight into agents, making them difficult to debug. As well as, developers often find themselves always switching between coding environments like GitHub Copilot and Cursor and the observability tools they’re using to maintain tabs on their code in production.

The startup’s solution is to treat agent sessions and huge language model conversations as standard telemetry signals, much like the logs, metrics and traces related to traditional applications. The thought is that users will give you the option to watch the performance of agents within the context of their broader information technology infrastructure.

The largest recent update is a brand new AI observability capability that’s launching as a public preview in Grafana Cloud today. It’s tasked with observing the behavior of AI agents, including their inputs, outputs and execution flows, in order that these may be constantly monitored for low-quality responses, policy violations and other anomalous activity. In keeping with Grafana, it’s in a position to surface risks comparable to data exposure or leaked credentials much faster than existing observability tools not designed for AI agents.

Developers will likely appreciate the brand new Grafana Cloud CLI, or GCX, command line tool, which is a brand new “agentic interface” that’s designed to live where they work. So relatively than having to leap from their integrated development environment to a Grafana dashboard, they will use GCX to invoke the Grafana Assistant directly inside environments like Claude Code or GitHub Copilot. The thought is to create a “continuous feedback cycle,” where developers can always watch their AI agents in production, correlate alerts and receive suggested code fixes in real time.

On the back end, Grafana says, it’s overhauling its core observability engine to higher cope with the large volumes of information generated by AI systems. The predominant focus here is Grafana’s log aggregation tool, which has been updated to Grafana Loki Evolution. Loki has been rebuilt from scratch on an Apache Kafka-based architecture and comes with a brand new query planner tool that accelerates performance by ten-times on aggregated queries while scanning 20 times less data.

Grafana can also be making its AI-powered Grafana Assistant tool available to more users. Previously limited to Grafana Cloud only, the AI sidekick is now being integrated with on-premises Grafana Enterprise deployments to cater to customers with strict data privacy requirements. It also gains recent features comparable to an “assistant workspace” for full-screen interactions and an “assistant API” that may extract Grafana’s insights into third-party tools.

Finally, in a nod to the open-source community, Grafana has published a brand new benchmarking tool called o11y-bench. It’s meant to measure how well AI agents perform various real-world tasks, comparable to fixing an app’s broken dashboard or investigating an outage, against a live Grafana stack. In keeping with the corporate, it’s meant to be a standardized “IQ test” that measures the effectiveness of AI observability agents.

Even though it went unsaid, today’s updates illustrate Grafana’s long-term vision. If AI agents are going to begin running enterprises at scale, they’ll need a supervisor. By forming a dedicated AI unit, tasked with “coordinating work across AI observability, assistant experiences and agent-driven workflows,” Grafana is set to tackle that role before another person steps up.

Image: SiliconANGLE/Microsoft Designer