Self-Hosting

Deploy your own Qiplim Studio instance on your infrastructure. The source code is not public yet — the open-core release is in preparation; contact us for early access.

Prerequisites

Runtime

Node.js 20+
pnpm 9.12+
Docker & Docker Compose

Services

PostgreSQL 16 (with pgvector)
Redis 7+
At least 1 AI API key (Mistral, OpenAI, Anthropic, or Google)

Installation

1. Get the source code (early access)

git clone REPO_URL_ACCES_ANTICIPE/studio.git   # contactez-nous pour l'acces anticipe
cd studio

2. Install dependencies

pnpm install

3. Start Docker services

The docker-compose includes PostgreSQL (with pgvector) and Redis:

pnpm docker:up

Services started: postgres-studio (port 5433), redis (port 6379).

4. Configure environment

cp apps/studio/.env.example apps/studio/.env
# Edit .env with your values

5. Sync database schema

pnpm db:push:studio

6. Start development server

pnpm dev:studio

The studio is accessible at http://localhost:3001.

Configuration (.env)

Main environment variables. See apps/studio/.env.example for the complete list.

Variable	Description	Example
`DATABASE_URL`	PostgreSQL connection URL	postgresql://qiplim:qiplim@localhost:5433/qiplim_studio
`REDIS_URL`	Redis connection URL	redis://localhost:6379
`NEXT_PUBLIC_APP_URL`	Public application URL	http://localhost:3001
`BETTER_AUTH_SECRET`	BetterAuth session secret	(openssl rand -hex 32)
`MISTRAL_API_KEY`	Mistral API key (optional)	sk-...
`OPENAI_API_KEY`	OpenAI API key (optional)	sk-...
`ANTHROPIC_API_KEY`	Anthropic API key (optional)	sk-ant-...
`GOOGLE_AI_API_KEY`	Google Gemini API key (optional)	AIza...

Docker Compose

The docker-compose.yml file at the project root defines the services:

services:
  postgres-studio:
    image: pgvector/pgvector:pg16
    environment:
      POSTGRES_USER: qiplim
      POSTGRES_PASSWORD: qiplim
      POSTGRES_DB: qiplim_studio
    ports:
      - "5433:5432"

  redis:
    image: redis:7-alpine
    ports:
      - "6379:6379"

Note: The pgvector/pgvector:pg16 image includes the pgvector extension for vector embeddings (RAG).

Production deployment

Production build

pnpm build:studio
# or
cd apps/studio && pnpm build

Clever Cloud deployment

Qiplim uses Clever Cloud (French hosting provider) in production. Required configuration:

Application: Node.js app with CC_NODE_VERSION=20
Database: PostgreSQL add-on with pgvector enabled
Cache: Redis add-on
Deployment: Push to the studio branch triggers automatic deployment

Other platforms

Qiplim Studio is a standard Next.js application. It can be deployed on any platform supporting Node.js: Vercel, Railway, Fly.io, AWS, etc. Make sure PostgreSQL with pgvector and Redis are accessible.

Troubleshooting

Database won't start

Check that Docker is running and port 5433 is not already in use. Use `pnpm docker:status` to check container status.

Error pgvector: extension not found

Make sure you're using the `pgvector/pgvector:pg16` image (not `postgres:16`). The pgvector extension is required for embeddings.

Prisma error after schema change

Re-run `pnpm db:push:studio` to sync the schema. If migration fails, delete and recreate the database: `pnpm docker:down && pnpm docker:up && pnpm db:push:studio`.

AI generations not working

Check that at least one AI API key is configured in `.env`. Keys are BYOK (Bring Your Own Key) -- the project does not include default keys.