Private Deployments
Run the complete Routero AI control plane inside your own infrastructure. Your VPC, your compute, your data — with zero dependency on Routero-managed systems after the initial deployment.
When to choose this
- Data never leaves your boundary — provider API keys, audit logs, and all operational data stay in infrastructure you own and control.
- Compliance requires it — FedRAMP, internal InfoSec mandates, customer contractual requirements, or air-gap environments that prohibit third-party-managed compute.
- Full upgrade control — you decide when to pull new images and roll forward; Routero Cloud’s continuous deployment is not forced on you.
- Custom network topology — lock the stack inside a private VPC, restrict egress to specific provider endpoints, integrate with your internal PKI.
Private Deployments are available on the Enterprise plan. Contact your solutions engineer to receive the deployment package, image registry access, and onboarding support.
What you get
The private deployment package ships the same stack that powers Routero Cloud:
| Component | What it does |
|---|---|
| Gateway proxy | OpenAI-compatible HTTP proxy — routing, policy, budgets, audit |
| Coworker service | Background worker for async tasks, cache warm-up, budget resets |
| Cerbos | PBAC/RBAC policy engine for fine-grained authorization |
| Admin console | Full dashboard — key management, teams, guardrails, prompt registry, memory sessions |
Your infrastructure provides the stateful layer: Postgres (primary datastore), Redis (caching and rate-limit counters), and optionally a vector store for Memory-as-a-Service.
Supported environments
| Platform | Notes |
|---|---|
| AWS | Reference architecture using ECS Fargate, RDS, ElastiCache. Full HA topology with multi-AZ. |
| Azure / GCP | Equivalent managed container + managed Postgres + Redis services. Topology guide available on request. |
| On-premises / air-gap | Kubernetes (or equivalent container runtime) + self-managed Postgres + Redis. Images mirrored to your internal registry. |
→ Reference Architecture for the canonical AWS topology (VPC · ALB · ECS Fargate · RDS · Redis · Cerbos).
Baseline infrastructure requirements
| Resource | Minimum | Recommended |
|---|---|---|
| Compute (proxy) | 2 vCPU / 4 GB RAM | Autoscaling group, 2+ replicas |
| Compute (coworker) | 1 vCPU / 2 GB RAM | 1–2 replicas |
| Postgres | db.t3.small · 20 GB | db.t3.medium · Multi-AZ |
| Redis | cache.t4g.small | cache.t4g.medium · Multi-AZ |
| Memory tier (optional) | Postgres + pgvector | + Neo4j for knowledge graph |
Estimated baseline cost (AWS): ~USD 300/month for a minimal production topology. Memory-tier services add ~USD 50–150/month.
Upgrades
Routero publishes updated container images on a regular release cadence. You control when to pull and deploy — there is no forced upgrade. Release notes are available in your customer portal.
Getting started
Contact solutions@routero.ai or your assigned solutions engineer. Onboarding includes:
- Image registry access and deployment package
- Architecture review call (infra sizing, network topology, compliance requirements)
- Deployment walkthrough and initial configuration
- Handoff to customer success for ongoing support