Akuity gives AI agents a governed path into production

Akuity Inc., a software delivery platform company, has launched Agentic Control Plane, a solution allowing AI agents to access pipeline data under existing platform permissions. The platform also introduced Akuity MCP Server, which connects agents to the platform via Model Context Protocol. Many AI agents are already operating in production environments without proper authorization, often connecting through open-source MCP servers that provide unrestricted access without identity or audit trails.

To address this, Akuity routes all agent requests through Agentic Control Plane, which authenticates the agent with the person's existing permissions, eliminating the need for separate credentials. A policy layer above this ensures sensitive production actions comply with regulations. The control plane provides an operational record of the pipeline, including deployment history, cluster health, and change lineage.

Engineers can ask a single question to an agent and receive a comprehensive answer across all clusters. The On-Call Agent handles more complex issues by analyzing logs and recent changes to suggest fixes. MLB is an early adopter, using the control plane to identify over 100 degraded applications within hours and trace a systemic issue to its root cause in under 10 minutes.

Michael Goodness, a principal DevOps engineer at MLB, praised the tool for its efficiency and potential for expanded adoption. Akuity co-founder and CEO Hong Wang emphasized that the next phase will focus on giving agents a reliable path to work on infrastructure changes, rather than adopting AI fastest. Akuity was founded by the creators of Argo CD and Kargo, with $20 million in Series A funding led by Lead Edge Capital in 2022.

AI automations have been on the platform since September, with on-call engineers no longer being paged for Kubernetes application incidents caught and fixed automatically.