Grafana's Big Tent Agents in Production: The Future of AI-Driven Observability
11 snips
Feb 27, 2026 Cyril Tovena, principal engineer who built the Grafana Assistant; Manoj Acharya, VP of Engineering with expertise in knowledge-graph observability; Spiros Xanthos, founder of Resolve AI working on agentic troubleshooting. They discuss agentic AI running alongside engineers. Topics include automated root cause analysis, knowledge and context graphs, agents crafting production queries, pricing and privacy tradeoffs, and trust-building for autonomous actions.
AI Snips
Chapters
Transcript
Episode notes
Agents Solve The Human Bottleneck In Observability
- Observability data is plentiful but human operators remain the bottleneck for reliability.
- Resolve builds agentic operators that use existing tools to troubleshoot, reducing time spent running and debugging large systems.
Databases Must Evolve For Agent First Access
- Existing observability databases were built for human use and need to evolve for agent access.
- Agents can overwhelm systems with naive queries, so tools must add agent-friendly APIs and throughput.
Provide Agents Purpose Built Tools Not Raw Query Access
- Give agents structured tools and abstractions for logs, metrics, code, and infra rather than raw access.
- Build two-layer abstractions so agents can generate precise queries across different backends without overwhelming them.