Deploy agenticos
AI agent platform for your company's AI teammates on your own servers
Just deployed
/app/media
Just deployed
Redis
Just deployed
/data
Just deployed
/var/lib/postgresql/data
Deploy and Host AgenticOS on Railway
AgenticOS is a self-hosted platform for building, running, and governing AI agents. It provides a web console, shared conversations, retrieval-augmented knowledge, MCP integrations, budgets, approvals, and audit history. Teams keep their data and provider credentials under their own control while configuring reusable agents through a browser-based workspace.
About Hosting AgenticOS on Railway
Railway runs the AgenticOS console and API from pinned GHCR images, with PostgreSQL and Redis as private supporting services. Railway provides managed networking, public HTTPS domains, persistent volumes, and straightforward environment-variable wiring between services.
Tech Stack
- FastAPI and Python backend
- Next.js and Bun frontend
- PostgreSQL with the pgvector extension
- Redis-compatible cache and queue service
- Pydantic AI agent runtime
- MCP integrations and retrieval-augmented generation
- Railway private networking, domains, and volumes
Why Deploy AgenticOS on Railway
Railway provides a fast path to a running self-hosted AI workspace without building the application locally. Pinned container images make releases reproducible, private service references keep database traffic off the public internet, and persistent volumes preserve PostgreSQL, Redis, and uploaded media data.
Common Use Cases
- Internal AI assistants for company teams
- Document question answering and knowledge search
- MCP-based access to SaaS tools
- Governed agents with budgets and approvals
- Shared chat and reusable agent workspaces
- API access to published agents
Deployment Notes
The public frontend listens on port 3000 and the backend API listens on port 8000. The frontend uses the app service's private Railway domain for server-side requests and the app's public domain for browser API and WebSocket traffic. PostgreSQL must retain its /var/lib/postgresql/data volume and includes the vector extension. Redis uses /data and password authentication. The app persists uploaded media at /app/media; after attaching a new Railway volume, ensure it is owned by UID/GID 1000 (appuser). The app startup runs migrations and an idempotent bootstrap using ADMIN_USERNAME and ADMIN_PASSWORD. Provider keys are added later through AgenticOS settings.
Dependencies for AgenticOS on Railway
AgenticOS requires PostgreSQL with pgvector and Redis for application state, sessions, caching, and background coordination. The frontend communicates with the backend over Railway private networking while users access the frontend through its public HTTPS domain.
Deployment Dependencies
| Service | Image | Port | Volume |
|---|---|---|---|
| app | ghcr.io/vstorm-co/agenticos-backend:1.89.1 | 8000 | /app/media |
| frontend | ghcr.io/vstorm-co/agenticos-frontend:1.89.1 | 3000 | - |
| Postgres | ghcr.io/railwayapp-templates/postgres-ssl:18 | 5432 | /var/lib/postgresql/data |
| Redis | redis:8.2 | 6379 | /data |
Template Content

