Deployment

Deploy Magnet AI to your environment. The guides below cover production-oriented setups and the operational concerns you’ll need to address (database, secrets, TLS/routing, and upgrades).

Container image: Magnet AI is published as a prebuilt container image: ghcr.io/ideaportriga/magnet-ai:latest.

Deployment options

• Docker Compose - Deploy Magnet AI in a production-like, single-host setup using Docker Compose (PostgreSQL + pgvector + Nginx reverse proxy).
• Kubernetes/OpenShift - Deploy Magnet AI to Kubernetes or OpenShift by running the Magnet AI container image and connecting it to a PostgreSQL database with pgvector enabled.
• Azure - Deploy Magnet AI on Azure using Azure Container Apps and Azure Database for PostgreSQL.

Database (required)

All deployment options require PostgreSQL with pgvector enabled.

• Docker Compose: includes a PostgreSQL + pgvector service in the stack (good for self-hosted single-host deployments). For higher availability, replace it with a managed PostgreSQL service and point DB_* variables to it.
• Kubernetes/OpenShift: this guide assumes you already have a PostgreSQL endpoint (managed DB or operator). Database installation is intentionally out of scope for the Kubernetes guide.
• Azure: the Azure guide includes provisioning Azure Database for PostgreSQL and enabling pgvector.

For initializing the database follow Database Initialization guide.

Minimum production checklist

• Secrets: store secrets in your platform secret manager; keep SECRET_ENCRYPTION_KEY stable across restarts.
• Routing/TLS: route /admin/, /panel/, /help/, /api/ and terminate TLS.
• Migrations: run Alembic migrations during deploy (containerized deployments typically set RUN_MIGRATIONS=true).