How to Deploy Rakazo with Docker Compose in 10 Minutes

Deploying Rakazo on your own server is straightforward thanks to the official docker-compose.prod.yml stack. In this tutorial you will get a production-ready instance running with Postgres, the Rakazo API, and your choice of LLM — in about ten minutes.

Prerequisites

You need a Linux server with Docker and Docker Compose installed. Rakazo recommends at least 4 GB of RAM and 2 CPU cores for a comfortable setup; plan for 8 GB if you run a local model alongside it.

Step 1 — Clone the repository

git clone https://github.com/elie222/rakazo.git
cd rakazo/infra/compose

Step 2 — Configure your environment

Copy the example environment file and set your secrets. The compose file expects POSTGRES_PASSWORD and will fail fast if it is missing — a nice safety touch.

cp .env.example .env
# edit .env: set POSTGRES_PASSWORD, your LLM API key, and RAKAZO_IMAGE_TAG

Step 3 — Start the stack

On a fresh checkout the default image tag is local, so the stack builds from source on the first run. Once you pin a released tag, deployments become a simple docker compose pull && up -d.

docker compose -f docker-compose.prod.yml up -d --build

Step 4 — Connect your model

Open the web app, sign in, and add your preferred LLM provider. Because Rakazo is model-agnostic, you can switch between providers per conversation — the sandbox stays isolated regardless of which model executes code.

Step 5 — Chat from anywhere

Install the desktop or mobile app and point it at your server. Your conversations sync to your own Postgres database, and the whole stack is backed up with the included backup-prod.sh script.

Troubleshooting tip: if the API container exits immediately, check that POSTGRES_PASSWORD is set and that the Postgres healthcheck passes (docker compose ps).