Skip to main content
Upon completion of this Quick Start process you’ll have a fully functional mAItion instance running locally and configured to use your S3 bucket as a data source.

​
Requirements

  • Docker and Docker Compose
  • S3 bucket with documents
  • OpenRouter or OpenAI API key

​
1. Prepare Configuration Files

Create .env.rag from .env.rag.example:
  • S3_ACCOUNT1_ENDPOINT: S3 API endpoint URL (https://s3.amazonaws.com for AWS S3)
  • S3_ACCOUNT1_ACCESS_KEY: access key ID used to read bucket objects
  • S3_ACCOUNT1_SECRET_KEY: secret key paired with the access key
  • S3_ACCOUNT1_REGION: bucket region (example: us-east-1)
  • S3_ACCOUNT1_BUCKETS: one bucket or comma-separated bucket names
  • S3_ACCOUNT1_SCHEDULES: ingestion interval in seconds (or comma-separated intervals, default is 3600)
  • S3_ACCOUNT1_USE_SSL: enable TLS for S3 connection (default is True)
  • OPENROUTER_API_KEY: API key for OpenRouter model calls (optional, only used for rephrase)
Create config.yaml from config.yaml.example. Default configuration is pre-configured for:
  • single S3 connector
  • sentence-transformers/all-mpnet-base-v2 for embeddings
  • openai/gpt-oss-20b:free for rephrase endpoint (optional, only used for rephrase)
Create .env from .env.openwebui.example:
  • set OPENAI_API_KEY
  • optionally set OPENAI_DEFAULT_MODEL (default is openai/gpt-4.1-nano)
  • optionally change HTTP_WEB_PORT (default is 3000)

​
2. Start Stack

Run: 
docker compose up -d
Wait for all services to become healthy: 
docker compose ps
It can take up to ~1 minute on cold start.

​
3. Login

Open http://localhost:3000. Use admin credentials from .env:
  • email: X_WEBUI_ADMIN_EMAIL
  • password: X_WEBUI_ADMIN_PASS
If unchanged, defaults are:
  • username: admin@example123.com
  • password: q1w2e3r4!