Giant Swarm, a Kubernetes specialist, has unveiled its Agent Platform, designed to operate AI agents as isolated, orchestrated workloads within an organization's existing Kubernetes infrastructure rather than as IDE extensions. The platform runs on-premises, in the cloud, at the edge, or in air-gapped networks without relying on proprietary cores or SaaS backends that transmit data externally. Giant Swarm positions the offering as sovereign, curated, and auditable, ensuring agents execute where data resides.

The company has operated its own fleet of agents on this infrastructure since mid-2025, deploying them for code reviews, incident triage, and backlog management. CEO Henning Lange announced the Agent Platform is now opening to selected partners and customers for testing, with a general public rollout planned for later in 2026.

Beyond enabling curated AI agent deployment in production environments, the platform addresses the growing problem of shadow AI within enterprises. According to an IDC study, approximately 56 percent of employees use unapproved AI tools compared to only 23 percent using organizational-provided tools. IBM research has shown that one in five organizations has suffered compromises through shadow AI systems, creating security and governance risks that structured platforms aim to mitigate.