Deployment

TurfAI is a set of cooperating services — the DMS/API layer (Strapi), a job router (Redis) and processors, an LLM service, the RAG query service, the event bus, and the document store (PostgreSQL + pgvector, plus GCS for files). It can be deployed as SaaS, into a customer VPC, or on-prem.

Service topology

The frontend and the MCP server talk only to the DMS (port 1338) — never directly to the Python services. The DMS routes jobs through Redis to the processors, which call the LLM/RAG services and integrations.

Operational references

The authoritative operational runbooks live in turfai-docs-source/operations/ and are synced into this guide. Don't duplicate env/secret values here — follow the runbooks:

• Deployment strategy — topology and rollout.
• Deployment checklist — go-live steps.
• Environment variables — the full inventory.
• Google OAuth setup — connecting Google for Drive/Gmail.
• Backup & restore — data recovery.
• Release 2.0 operator runbook — release operations.