Mirantis Launches Agent Platform for Governed AI Agents on Kubernetes

SALT LAKE CITY -- At KubeCon + CloudNativeCon North America 2026, Mirantis unveiled the general availability of its Mirantis Agent Platform. This self-hosted, governed environment enables the execution of AI agents in production on Kubernetes. Mirantis is complementing the platform with Forward Deployed Engineering, a service where senior Mirantis engineers collaborate closely with customer teams to design, integrate, and operationalize custom agents before transferring control to the customer.

The platform caters to platform engineering, site reliability engineering (SRE), and IT operations teams who seek control over agent access, spending limits, and agent action tracking while maintaining human oversight over operational fixes.

Key features of the Mirantis Agent Platform encompass policy-based governance with sandboxed isolation for individual agents, adaptable spending caps at the organization, project, and agent levels, comprehensive audit trails of every agent action, a one-click mechanism to restrict network access for all agents simultaneously, and human-approved remediation.

The platform supports Mirantis' native Prism agent and third-party agents that customers already utilize. For infrastructure operations, it identifies production problems, determines root causes, and proposes fixes, which an agent implements only after obtaining human approval.

Mirantis' Vice President of Product Management, Ronen Kofman, highlighted the importance of enterprise AI adoption, stating that it is faltering not due to a lack of model intelligence but because organizations lack governance and execution support. By integrating sandboxed execution and real-time human oversight into the Mirantis Agent Platform, coupled with Mirantis Forward Deployed Engineering, Mirantis aims to provide the necessary guardrails for organizations to safely execute AI agents in production.