DeepSynth is a web-based chatbot powered by a local LLM (Google's DeepMind Gemma 3 4B, distilled 4-bit quantized model). It allows users to chat locally with the model and download conversation logs as JSON.
No database is required. Conversations are handled in memory and can be exported directly from the frontend.
- 🧠 Local LLM Integration using Google's DeepMind Gemma 3 4B
- 💬 Interactive Chat Interface built with React and Tailwind CSS
- 🔒 Authentication powered by Auth0 SPA + JWT validation
- 📥 Downloadable Chat Logs exported as structured JSON files
- ⚡ Node.js + Express Backend for API routing and model communication
- 🐳 Dockerized Deployment with Docker Compose
- 🚫 No Database Required
- React.js
- Tailwind CSS
- Auth0 SPA SDK
- Node.js
- Express.js
- JWT Authentication
- Ollama
- Google DeepMind Gemma 3 4B (4-bit quantized)
- Docker
- Docker Compose
- Docker
- Docker Compose
- Ollama installed locally
- Gemma model pulled via Ollama
ollama pull gemma3:4bdocker compose up --builddocker compose up -d --builddocker compose downgit clone https://github.com/render-thevoid/deepsynth-ai.git
cd deepsynth-aicd client
npm install
npm run devcd server
npm install
npm startAUTH0_AUDIENCE=your_audience
AUTH0_DOMAIN=your_domain
PORT=5000VITE_AUTH0_DOMAIN=your_domain
VITE_AUTH0_CLIENT_ID=your_client_id
VITE_API_URL=http://localhost:5000.
├── client/
│ ├── src/
│ ├── Dockerfile
│ └── package.json
│
├── server/
│ ├── src/
│ ├── Dockerfile
│ └── package.json
│
├── docker-compose.yml
└── README.md
- Light mode
- Persistent chat history with a database
- Multiple model selection
- Enhanced analytics for exported conversations
- Mobile-first responsive design
- Streaming responses from the model
DeepSynth runs locally and keeps your conversations private.
Unlike most modern software, it does not immediately try to sell your thoughts to an advertising algorithm.



