A complete, self-hosted private AI stack in a single docker compose up -d.
Run your own ChatGPT, RAG pipelines, private search, and AI workflows — entirely on your own hardware. No cloud APIs, no subscriptions, no data leaving your network.
| Service | What it does |
|---|---|
| Ollama | Run LLMs locally (Llama 3, Mistral, Gemma, etc.) |
| Open WebUI | ChatGPT-like interface for your local models |
| ChromaDB | Vector database — chat with your own documents |
| n8n | Visual workflow automation connecting everything |
| SearXNG | Private meta-search (Google, Bing, DuckDuckGo — no tracking) |
| PostgreSQL | Reliable database backend for n8n |
Optional extras (one flag to enable):
- LibreTranslate — ML-powered translation API
- Redis — Caching layer for API responses
- ChromaDB Admin — Visual management for your vector database
# 1. Configure
cp .env.example .env
nano .env # change POSTGRES_PASSWORD at minimum
# 2. Launch
docker compose up -d
# 3. Pull an AI model
./scripts/pull-models.sh llama3.2:3b # ~2 GB download, 10-40 min
# 4. Use it
# Open WebUI: http://localhost:3000
# n8n: http://localhost:5678
# SearXNG: http://localhost:80805 pre-built n8n workflows ready to import:
- RAG Chat with Documents — Upload a PDF, ask questions, get answers citing your document
- Private Web Search — Search the web via SearXNG, get AI-summarized results
- Knowledge Base Ingest — Auto-embed documents into ChromaDB for later querying
- Web Scrape & Summarize — Give it a URL, get a clean AI summary back
- Translation Pipeline — Text in, translated text out (requires LibreTranslate)
See docs/WORKFLOWS.md for import instructions and usage.
Everything starts with one command. Optional services via --profile extras.
docker compose up -d # Core services
docker compose --profile extras up -d # + LibreTranslate, Redis, ChromaDB AdminRun only the stacks you need. Each is independent.
docker network create homeai-net # Shared network (once)
cd stacks/llm && docker compose up -d # Just Ollama + Open WebUI
cd stacks/rag && docker compose up -d # Just ChromaDB
cd stacks/automation && docker compose up -d # Just n8nSee the stacks/ directory for all options.
Running on a Synology NAS? See synology/SYNOLOGY-GUIDE.md for DSM-specific setup.
| Setup | RAM | CPU | Storage |
|---|---|---|---|
| Minimum | 8 GB | 4 cores | 20 GB |
| Recommended | 16 GB | 4+ cores | 50 GB |
| Full (13B+ models) | 32 GB+ | 8+ cores | 100 GB+ |
See docs/HARDWARE.md for details and Synology model compatibility.
- SETUP.md — Step-by-step setup checklist
- HARDWARE.md — Hardware compatibility matrix
- WORKFLOWS.md — Workflow import and usage guide
- TROUBLESHOOTING.md — Common issues and fixes
- LICENSES.md — Open source licenses for all services
- SYNOLOGY-GUIDE.md — Synology NAS setup
Questions? Issues? Visit http://linkedbits.net and leave a message