Rent the agents
Your logic and your users routed through a vendor's cloud control plane.Telemetry funnels back into their platform.Lock-in decided before anyone signs.
OMNIA · THE SELF-HOSTABLE AGENTOS
Build your agent platform in-house.Own the agents, rent the model.
The operating system for production agents — self-hosted on the open components your platform team already runs: Kubernetes, OCI, MCP, OpenTelemetry, Postgres. Author, prove, run, measure, and ship agents as installable bundles. Your cluster, your data, no proprietary control plane.
~ ❯ helm install omnia oci://ghcr.io/altairalabs/charts/omnia
# install the operator into your cluster
~ ❯ kubectl apply -f agent.yaml
agentruntime/checkout-agent created
✓ reconciled · replicas 2/2 · drift none
facade ws :8080 · sessions pg+redis · p50 42ms
# your cluster. your data. no control plane. 2/2 reconcileddrift nonep50 42ms · scale-to-zero
THE BET
Own the agents. Rent the model.
Serious enterprises never put core banking, ERP, or the trade engine on someone else's SaaS — they build the core in-house and wire in cloud products at the edges. The agent layer is becoming exactly that kind of core system: it orchestrates your proprietary business logic and sits between you and your users.
The model is a commodity — rent it, self-host an open one, or grow your own. Theloop — user ↔ agent ↔ your logic — you keep. That's the IP no serious enterprise cedes to a vendor's cloud.
Own them, in your cluster
Runs on the stack you already operate — K8s, OCI, OTel, Postgres.Open source — pull it, read it, run it, before you sign.Your agents and telemetry stay yours.
CAPABILITIES
A Kubernetes operator for the whole agent lifecycle.
Deploy agents as custom resources, scale them to zero, wire in tools, and watch them like any other workload — all with the observability stack you already run.
Kubernetes-native
Deploy agents as custom resources — AgentRuntime, PromptPack, Provider — with full GitOps support.
Any LLM provider
Claude, OpenAI, and Gemini behind one Provider resource — switch models without touching the agent.
Autoscaling
Scale-to-zero with KEDA, or standard HPA down to a warm floor — driven by active connections. You pay for load, not idle.
Tools & MCP
HTTP, gRPC, and MCP tool adapters register through a ToolRegistry — tools are just K8s microservices.
Sessions & memory
Postgres-backed sessions with a Redis warm cache stay persistent across replicas and restarts.
Observability
Prometheus metrics, Grafana dashboards, Loki logs, Tempo traces — integrated, not bolted on.
Built on what you already runKubernetesOCIMCPOpenTelemetryPostgresRedisKEDA
FIND YOUR PATH
Documentation, four ways.
The Omnia docs follow Diátaxis — tutorials to learn, how-to guides to get things done, reference to look things up, explanation to understand the why.
PRACTICAL STEPSTHEORETICAL KNOWLEDGEWHEN LEARNING
Learning-oriented
Tutorials
From helm install to your first agent answering over a WebSocket — no prior setup assumed.
Understanding-oriented
Explanation
The thinking behind the platform — the why, not the how.
Task-oriented
How-to guides
Focused recipes for the operational jobs you already know you need to do.
Information-oriented
Reference
Every custom resource, field, and option — exact and complete, for looking up.
INSTALL
One Helm chart into your cluster.
Install the operator over OCI, apply an AgentRuntime, and your agent reconciles into a running pod — facade, session wiring, and autoscaling. Nothing in here you don't already operate.
$ helm install omnia oci://ghcr.io/altairalabs/charts/omniaapiVersion: omnia.altairalabs.ai/v1alpha1
kind: AgentRuntime
metadata:
name: checkout-agent
spec:
promptPackRef: {name: my-pack}
providers:
- name: default
providerRef: {name: claude-provider}
facades: [{type: websocket, port: 8080}]requires kubernetes 1.28+ · helm 3.x · apache-2.0