Giant Swarm has opened its Kubernetes-native AI agent platform to customers, enabling organisations to run open-source AI agents entirely within their own infrastructure while retaining full data control and eliminating SaaS dependency.
Giant Swarm has opened its Kubernetes-native AI Agent Platform to selected partners and customers, bringing a production-tested platform built around open-source components and sovereign AI deployment to enterprises. A broader rollout is planned later this year.
Unlike SaaS-based AI services, the platform runs AI agents as isolated workloads inside customers’ own Kubernetes environments, supporting on-premises, air-gapped, edge, cloud and hybrid deployments. Giant Swarm said there is “no proprietary core or SaaS that phones home with the data.”
The platform’s “sovereign” design relies on open-source, interchangeable models, tools and harnesses, giving organisations full control over data while enabling auditable AI operations. It is built on Kubernetes, Cluster API, Flux Operator for GitOps, Backstage and the Grafana LGTM observability stack, with Kubernetes RBAC, Network Policies and SSO integration providing enterprise security.
Giant Swarm said the platform converts recurring agent runs into deterministic MCP workflows. Based on internal production use since mid-2025, it reduced cost per agent run by 2.8×, cut tool calls by 17×, supported around 500 parallel agents, and helped increase monthly pull requests by more than 300% year over year without adding headcount.
As CEO Henning Lange noted, the company has been operating its fleet of agents on the same platform infrastructure since mid-2025 for tasks including “code reviews, incident triage, and backlog processing.” The launch reflects a broader industry shift towards orchestrating fleets of AI agents on open-source cloud-native infrastructure rather than relying on standalone coding assistants.
