Grafana Labs Unleashes AI for System Operations

Grafana Labs is enhancing its Grafana Cloud platform with new agentic operations and AI capabilities. These updates address the dual challenge of using AI agents to manage systems and gaining visibility into these agents' behavior.

New features include Assistant Watchers, which are always-on agents designed to detect changes not covered by standard alerts by learning normal system patterns. Grafana Managed Sandboxes are also introduced, providing a secure cloud environment for AI agents to develop and test fixes using code and live telemetry, with proposed changes submitted via pull requests for human review.

Furthermore, Grafana Cloud Agent Observability now includes OpenTelemetry support. This integration ensures that agent activity is visible using the same open standard applied to other services, offering teams a unified view of their AI agents' performance and cost without needing separate tools.

These developments aim to increase operational efficiency, with past research indicating potential reductions in mean time to resolution and on-call shift burden. The company is focused on building a platform where agents can act with human oversight, allowing organizations to tailor autonomy based on their risk tolerance.