Grafana Assistant Now Pre-Loads Infrastructure Knowledge, Slashing Incident Response Time

Breaking: Grafana Assistant Pre-Builds Infrastructure Knowledge Base

Grafana has unveiled a major upgrade to its AI-powered Assistant, enabling it to automatically learn an organization's infrastructure before any query is made. The new capability eliminates the need for engineers to share context during incidents, cutting response times significantly.

“Assistant no longer starts from scratch. It studies your environment in the background and builds a persistent knowledge base of services, dependencies, metrics, and logs,” said Maria Chen, Senior Product Manager at Grafana Labs. “When an alert fires, the assistant already knows where to look.”

How It Works

The system runs silently across Grafana Cloud stacks. A swarm of AI agents performs four key tasks:

• Data source discovery – Identifies all connected Prometheus, Loki, and Tempo sources.
• Metrics scans – Queries Prometheus data in parallel to find services, deployments, and components.
• Enrichments via logs and traces – Correlates Loki and Tempo data with metrics to understand log formats, trace structures, and dependencies.
• Structured knowledge generation – Produces documentation per service group: identity, key metrics, deployment details, dependencies, and more.

All this happens with zero configuration, according to Grafana. The knowledge base updates continuously as infrastructure changes.

Background

Traditionally, AI assistants require engineers to manually share context about data sources, service connections, and relevant labels. This discovery process consumes precious minutes during incidents, when speed is critical.

“Every conversation used to start from scratch. Engineers had to explain their entire setup before getting useful insights,” noted Dev Patel, an SRE at a large e-commerce firm using early-access Grafana Assistant. “Now the assistant already knows my payment system talks to three downstream services and where its latency metrics live.”

Grafana Assistant first launched as an agentic observability tool earlier this year. The new pre-loading feature turns it into a proactive troubleshooting partner rather than a reactive Q&A bot.

What This Means

For incident response teams, the upgrade can shave off minutes from mean time to resolution (MTTR). An engineer can ask “why is my checkout service slow?” and immediately get answers tied to pre-known dependencies and data sources.

“This is a game-changer for teams where not everyone knows the full infrastructure,” said Chen. “A developer investigating their own service can now get accurate information about upstream dependencies they’ve never touched.”

The feature also reduces cognitive load during high-stress incidents. Instead of digging through dashboards or asking colleagues, engineers can rely on Assistant to have the latest map of the environment.

Grafana plans to extend the knowledge base to support more data sources and deeper integration with incident response workflows in future releases.