Quick Start
Get OfficePlane running locally in under 5 minutes using Docker.
Prerequisites
- Docker and Docker Compose
- A Google Gemini API key (for document ingestion)
1. Clone and Configure
git clone https://github.com/officeplane/officeplane
cd officeplane
Create a .env file in the project root:
GOOGLE_API_KEY=your-gemini-api-key
OPENAI_API_KEY=your-openai-api-key # optional, for content generation
2. Start Services
docker compose up -d
This starts four services:
| Service | URL | Purpose |
|---|---|---|
| UI | http://localhost:3000 | Next.js frontend dashboard |
| API | http://localhost:8001 | FastAPI backend |
| PostgreSQL | localhost:5433 | Document storage + pgvector |
| Redis | localhost:6379 | Task queue + pub/sub |
3. Install Pillow (first time only)
docker exec officeplane-api pip install Pillow
4. Verify
# Check API health
curl http://localhost:8001/health
# Open the dashboard
open http://localhost:3000
5. Upload Your First Document
curl -X POST http://localhost:8001/api/documents/upload \
-F "file=@path/to/your-document.docx"
Or use the Upload button in the UI dashboard.
The ingestion pipeline will:
- Detect the file format (DOCX/PDF)
- Convert to PDF if needed (via LibreOffice)
- Render each page to an image
- Send images to Gemini vision for structure extraction
- Parse the result into a document hierarchy
- Store everything in PostgreSQL with vector embeddings
What's Next
- Upload your first document — Full walkthrough with screenshots
- Generate a presentation — Create content from a prompt
- Configuration — Tune environment variables