Breaking News: Grafana Labs today announced a major update to its AI-powered observability tool, Grafana Assistant. The assistant now automatically builds a persistent knowledge base of your infrastructure, eliminating the need to manually share context during incident investigations.
When an unexpected alert fires, engineers typically spend critical minutes explaining their environment to AI assistants. With the new update, Assistant already knows which services are running, how they connect, where metrics live, and what logs look like—before the first question is asked.
“This is a fundamental shift in how teams approach incident response,” said Dr. Lisa Chen, Senior Product Manager at Grafana Labs. “Instead of wasting precious time teaching the tool about your stack, Assistant has already done its homework. It’s like having a team member who invisibly scouts the entire system 24/7.”
Background
Traditional AI assistants for observability require users to provide data source details, label mappings, and service dependencies each time a query is made. This manual “context sharing” eats into troubleshooting time exactly when speed matters most.
Grafana Assistant runs a background process that identifies all connected Prometheus, Loki, and Tempo data sources in a Grafana Cloud stack. It then scans metrics in parallel to discover services, deployments, and infrastructure components. Logs and traces are correlated to add context such as log formats, trace structures, and service dependencies.
How It Works – Zero Configuration
“Assistant operates with zero configuration,” explained Raj Patel, Lead Engineer on the Assistant team. “A swarm of AI agents automatically discovers data sources, fingerprints metrics, and cross-references logs and traces. By the time an engineer asks their first question, the assistant already holds a detailed map of the environment.”
- Data source discovery: Identifies Prometheus, Loki, and Tempo sources in your stack.
- Metrics scans: Queries via parallel agents to find services, deployments, and components.
- Enrichments: Correlates logs and traces with metrics for deeper context.
- Structured knowledge: Generates documentation per service group: what it is, key metrics, deployment details, dependencies.
The result is a persistent knowledge base that grows and updates automatically. Engineers can ask detailed questions without preamble—for example, “What are the upstream dependencies of the payment service and their current latency?”—and get precise, instant answers.
What This Means
For incident response teams, every second counts. Preloading infrastructure context can shave valuable minutes off mean time to resolution (MTTR). This is especially powerful for organizations with complex microservice architectures where no single person has the full picture.
“A developer who has never touched the billing system can now ask Assistant about its health and dependencies,” said Dr. Chen. “This democratizes observability—you don’t need deep system knowledge to start troubleshooting effectively.”
The feature also reduces cognitive load during high-stress events. Instead of paging through documentation or Slack histories, engineers get context instantly. Grafana emphasizes that the knowledge base remains private and is only accessible within the user’s own Grafana Cloud instance.
To see how Assistant can transform your incident response, try it now or explore the Background section above for more technical details.