🏗️ FlowTech-AI - Final Architecture

Last updated: 2025-10-19

🎯 What is FlowTech-AI?

FlowTech-AI is a production-ready, self-hosted AI stack that provides:

✅ Cursor AI Enhancement (MCP-Qdrant)
✅ Conversational AI (OpenWebUI with RAG)
✅ Automation Platform (n8n)
✅ Vector Database (Qdrant)
✅ One-command deployment (./init.sh)

Status: ✅ Production-ready, open-source, fork-friendly

📦 What's Included

Core Services (docker-compose.yml)

Service	Port	Description	Status
OpenWebUI	8081	AI chat with RAG	✅ Production
MCP-Qdrant	8000	Cursor integration	✅ Production
n8n	5678	Workflow automation	✅ Production
Qdrant	6333	Vector database	✅ Production
PostgreSQL	5432	Metadata storage	✅ Production
Redis	6379	Cache & queues	✅ Production
SearxNG	8082	Web search	✅ Production
Langfuse	3300	LLM observability	✅ Production
ClickHouse	8123	Analytics	✅ Production
MinIO	9000	S3-compatible storage	✅ Production

Templates & Examples

Notes/_Templates/ - Generic Obsidian templates for note organization
SRC/ - Example n8n workflows
cursor-mcp-config.json - Cursor configuration template

Documentation

README.md - Main documentation
QUICKSTART.md - 5-minute setup guide
docs/ - Technical documentation
Notes/README-NOTES.md - Templates guide

🔄 Data Flow

1. Cursor → MCP-Qdrant → Qdrant

Developer codes in Cursor
  ↓
@qdrant store "code snippet"
  ↓
MCP-Qdrant (port 8000)
  ↓
Qdrant collection: cursor-context
  ↓
@qdrant find "snippet"
  ↓
AI retrieves context

Use case: Store and retrieve code context during development

2. OpenWebUI → Qdrant → RAG

User uploads documents in OpenWebUI
  ↓
OpenWebUI processes & embeds (bge-m3:567m)
  ↓
Qdrant collection: open-webui_files
  ↓
User asks question
  ↓
RAG retrieves relevant chunks
  ↓
LLM generates answer with context

Use case: Q&A over technical documentation

3. n8n → External APIs

Trigger (webhook, schedule, manual)
  ↓
n8n workflow execution
  ↓
API calls, data processing
  ↓
Results stored or notifications sent

Use case: Automated workflows, integrations

🛠️ Technology Stack

AI & LLM

Ollama (external): http://YOUR_OLLAMA_HOST:11434 (or localhost if running locally)
Embedding model: bge-m3:567m (1024 dimensions, multilingual)
Vector DB: Qdrant 1.11.5

Backend

PostgreSQL: Metadata, n8n workflows, Langfuse traces
Redis: Caching, session management
ClickHouse: Analytics, telemetry
MinIO: S3-compatible object storage

Frontend

OpenWebUI: React-based chat interface
n8n: Node.js workflow editor
Langfuse: Next.js observability UI

Integration

MCP-Qdrant: Python FastAPI server for Cursor
SearxNG: Metasearch engine for web queries

📁 Directory Structure

FlowTech-AI/
├── docker-compose.yml           # All services configuration
├── init.sh                      # One-command deployment
├── .env                         # Configuration (generated by init.sh)
│
├── AI_Data/                     # Runtime data (Docker volumes)
│   ├── openwebui/               # OpenWebUI data
│   ├── qdrant/                  # Vector database storage
│   ├── n8n/                     # Workflows & credentials
│   ├── mcp-qdrant/              # MCP service data
│   └── ...
│
├── Notes/
│   ├── _Templates/              # Generic Obsidian templates
│   └── README-NOTES.md          # Templates guide
│
├── mcp-qdrant/                  # MCP service Dockerfile
│   ├── Dockerfile
│   └── README.md
│
├── SRC/                         # Example workflows
│   └── FlowTech-AI-Complete-Workflow.json
│
├── docs/                        # Technical documentation
│   ├── MCP-QDRANT.md
│   ├── DEVELOPER_GUIDE.md
│   └── ...
│
├── README.md                    # Main documentation
├── QUICKSTART.md                # Setup guide
└── cursor-mcp-config.json       # Cursor config template

🔒 What's NOT Included (Private)

❌ Personal notes (VMs, Servers, Domains)
❌ Automated notes sync scripts
❌ Samba network shares
❌ Personal credentials & secrets

For personal notes management: See Flow-Notes-AI (private companion repo)

🚀 Deployment

Minimum Requirements

OS: Ubuntu 20.04+ / Debian 11+ / RHEL 8+
CPU: 4 cores
RAM: 16 GB
Disk: 50 GB
Docker: 24.0.0+
Docker Compose: 2.20.0+

Quick Deploy

git clone https://github.com/FlowTech-Lab/FlowTech-AI.git
cd FlowTech-AI
sudo ./init.sh

That's it! All services start automatically with health checks.

🔧 Configuration

Environment Variables (.env)

Auto-generated by init.sh, key variables:

# External Ollama
OLLAMA_BASE_URL=http://localhost:11434

# Ports
OPENWEBUI_PORT=8081
MCP_QDRANT_PORT=8000
N8N_PORT=5678

# Security (auto-generated)
POSTGRES_PASSWORD=<random>
REDIS_AUTH=<random>
N8N_BASIC_AUTH_PASSWORD=<random>

# RAG Configuration
RAG_EMBEDDING_MODEL=bge-m3:567m
RAG_EMBEDDING_ENGINE=ollama
VECTOR_DB=qdrant

Cursor Configuration (~/.cursor/mcp.json)

{
  "mcpServers": {
    "qdrant": {
      "transport": "sse",
      "url": "http://YOUR_SERVER_IP:8000/sse",
      "timeout": 30000
    }
  }
}

📊 Monitoring

Health Checks

# All services status
docker compose ps

# Service logs
docker compose logs -f openwebui
docker compose logs -f mcp-qdrant

# Qdrant collections
curl http://localhost:6333/collections

Langfuse Observability

URL: http://localhost:3300
Tracks: LLM calls, latencies, costs
Integration: n8n workflows, OpenWebUI

🔄 Updates

Pull Latest Changes

cd FlowTech-AI
git pull origin main
docker compose pull
docker compose up -d

Rebuild Services

docker compose build mcp-qdrant
docker compose up -d mcp-qdrant

🤝 Contributing

Want to improve FlowTech-AI?

Fork the repo
Create feature branch: git checkout -b feature/amazing-feature
Test changes: ./init.sh
Commit: git commit -m 'Add amazing feature'
Push: git push origin feature/amazing-feature
Open Pull Request

📚 Use Cases

1. AI-Enhanced Development (Cursor)

Store code snippets with @qdrant store
Retrieve context with @qdrant find
Build personal code knowledge base

2. Technical Documentation Q&A (OpenWebUI)

Upload technical docs (PDF, MD, DOCX)
Ask questions in natural language
Get accurate answers with citations

3. Workflow Automation (n8n)

Automate repetitive tasks
Integrate APIs and services
Schedule jobs and webhooks

4. Team Knowledge Base

Share OpenWebUI instance
Common documentation repository
Collaborative RAG

🆘 Support

Issues: https://github.com/FlowTech-Lab/FlowTech-AI/issues
Discussions: https://github.com/FlowTech-Lab/FlowTech-AI/discussions
Documentation: docs/ folder

📄 License

MIT License - See LICENSE

🔗 Related Projects

Flow-Notes-AI - Personal notes sync (private repo)
OpenWebUI - AI chat interface
n8n - Workflow automation
Qdrant - Vector database
Cursor - AI-powered IDE

Made with ❤️ by the FlowTech-Lab community

FilesExpand file tree

ARCHITECTURE-FINAL.md

Latest commit

History