Grafana Labs this week made generally available an artificial intelligence (AI) agent, dubbed Grafana Assistant, for its namesake dashboard in addition to previewing Grafana Assistant Investigations, an AI incident management tool that analyzes the observability stack, generates findings and hypotheses, and surfaces actionable recommendations for mitigation and remediation.
Announced at its ObservabilityCON 2025 conference, Grafana Assistant makes it possible to use natural language to generate and refine queries in PromQL, LogQL, and TraceQL in addition to now including support for specific roles and permissions to provide more granular access controls.
Additionally, Grafana Labs is making generally available a customizable Adaptive Traces tool and Grafana Federal Cloud with FedRAMP High and DoD IL5 compliance while at the same previewing Adaptive Profiles, a tool that enables IT teams to dynamically capture the most valuable performance data from production workloads in a way that provides more granular control over storage costs.
Grafana Labs also unfurled a Bring Your Own Cloud (BYOC) option through which it will, for a flat fee, manage instances of its observability platform on cloud infrastructure preferred by an internal IT team.
The company is also previewing an extension of its observability platform that adds support for databases, in addition to adding a Root Cause Analysis Workbench (RCA Workbench) that consolidates anomalies, dependencies and timelines into a single view. Integrated with Grafana Assistant, the RCA Workbench promises to reduce the time required to investigate incidents to a matter of minutes by leveraging the Grafana Cloud Knowledge Graph, formerly known as Asserts, to integrate metrics, logs, traces and profiles across a full stack of IT platforms.
Company CTO Tom Wilkie said collectively these capabilities will make it simpler for IT professionals of any skill level to query telemetry data to determine the root cause of a specific issue in a way that better identifies high-value telemetry data that is worth the cost of storing. An Entity Catalog, meanwhile, automatically discovers and maps all services and dependencies, surfacing curated dashboards with real-time health insights without requiring deep PromQL expertise.
Longer term, Grafana Labs also expects to be able to reach beyond OpenTelemetry to collect data by leveraging AI to convert legacy proprietary data monitoring formats into an OpenTelemetry-compatible format, added Wilkie.
As IT environments become increasingly more complex, the need for observability platforms that enable IT teams to go beyond merely monitoring a set of pre-defined metrics has become all too apparent. Historically, the challenge has been the level of programming expertise required to derive value from these tools, an issue that is now being addressed by AI tools that eliminate that requirement, noted Wilkie.
Of course, discovering the root cause of an issue is not the same as remediating it. Most IT teams are still far away from using AI agents to self-heal IT environments, but that goal might be achievable, said Wilkie.
In the meantime, however, there is little doubt that AI is about to democratize observability to the point where just about any IT professional can now investigate any anomalous activity discovered.
