
Deploy memoh
Every agent gets its own computer, desktop, network, and long-term memory
postgres
Just deployed
/var/lib/postgresql
Just deployed
/opt/memoh/data
pgvector
Just deployed
/var/lib/postgresql
Just deployed
Deploy and Host Memoh on Railway
Memoh is an open-source, multi-member AI agent platform for building and operating persistent bots. It combines a Vue web console, Go agent runtime, PostgreSQL-backed memory, provider configuration, channel integrations, and isolated workspaces. This Railway deployment provides a ready-to-use browser interface while keeping application data and credentials under your project's control.
About Hosting Memoh on Railway
Railway runs Memoh as Docker image services with a public web entrypoint, a private Go server, and persistent PostgreSQL services. Railway private networking connects the web proxy to the server, while generated variables provide secure initial credentials and service-to-service database wiring.
Tech Stack
- Vue 3 and Vite web UI
- Go and Echo API with the in-process AI agent runtime
- PostgreSQL with pgvector support
- Docker-based isolated workspace runtime
- Railway private networking, domains, variables, and persistent volumes
Why Deploy Memoh on Railway
Railway provides one-click service deployment, managed public routing, private service DNS, generated secrets, persistent volumes, and straightforward environment configuration. The template exposes the management UI immediately and keeps the API and databases private to the project network.
Common Use Cases
- Configure AI providers and create persistent multi-member bots
- Manage channels, tools, MCP integrations, and agent credentials
- Operate self-hosted AI workspaces with PostgreSQL-backed memory
- Prototype an internal AI assistant platform from a browser
Deployment Notes
- Open the generated web domain over HTTPS; the web service listens on container port 8082 via PORT=8082.
- The private server exposes the API on port 8080 and channel webhooks on port 8081; its bundled local Nginx entrypoint listens on 8082.
- Initial login credentials are supplied through ADMIN_USERNAME and the generated ADMIN_PASSWORD template variable.
- PostgreSQL and pgvector use persistent volumes. Configure an LLM provider in the Memoh UI before creating agent turns.
- No startCommand override is required; the Docker image entrypoints perform startup and migrations.
- Workspace and container features may require the Railway runtime capabilities available to the selected deployment.
Dependencies for Memoh on Railway
Memoh uses a web UI and private server service backed by PostgreSQL. The pgvector service is provisioned as the vector database dependency for semantic memory features.
Deployment Dependencies
| Service | Image | Port | Volume |
|---|---|---|---|
| web | memohai/web:0.20.0 | 8082 | - |
| server | ***********/memoh-railway:latest | 8082 | /opt/memoh/data |
| postgres | postgres:18.6-alpine | 5432 | /var/lib/postgresql |
| pgvector | pgvector/pgvector:0.8.6-pg18 | 5432 | /var/lib/postgresql |
Template Content
postgres
postgres:18.6-alpinepgvector
pgvector/pgvector:0.8.6-pg18