The Intelligent AI Agent Builder for Everyone.
AutoFlow AI is a powerful, low-code platform that allows you to build, visualize, and deploy AI automation agents effortlessly. It combines a state-of-the-art React-based Visual Builder with a robust Node.js + Google Gemini AI Backend.
- 🧠 Google Gemini Pro Integration: Powered by the latest Gemini 1.5 Flash models for lightning-fast and cost-effective AI reasoning.
- 🎨 Professional Visual Builder:
- Dark Mode infinite canvas.
- Agent-Centric Topology: All workflows centralized around your "AutoFlow Agent".
- Dynamic Bezier Curves: Smooth, professional connection lines.
- 🔌 100+ Integration Tools: Pre-built nodes for WhatsApp, Gmail, Slack, Salesforce, Stripe, and more.
- 🛠️ Customization Mode: Toggle between simple AI generation and advanced "Tech Mode" for manual fine-tuning.
- 💬 Natural Language to Workflow: Just type "Create a WhatsApp bot for order tracking" and watch the AI build it for you.
This is a monorepo containing both the frontend and backend:
```
AutoFlow/
├── backend/ # Node.js Express Server + Gemini AI Logic
│ ├── src/
│ ├── server.js
│ └── .env (Required)
│
└── autoflow-frontend/ # React + Vite + Tailwind Builder
├── src/
└── package.json
```
- Node.js (v18 or higher)
- Google AI Studio API Key (Free Tier supported)
```bash cd backend npm install ```
Create a .env file in the backend folder:
```env
GEMINI_API_KEY=your_google_ai_studio_key_here
PORT=3000
```
Start the server: ```bash node server.js ``` > Server runs on http://localhost:3000
Open a new terminal: ```bash cd autoflow-frontend npm install npm run dev ``` > Builder runs on http://localhost:5173
- Open the Builder in your browser.
- Chat with AI: Type a prompt like "Build a customer support agent that handles refunds via email".
- Visualize: The AI will instantly generate a star-topology workflow centered around your AutoFlow Agent.
- Customize:
- Click "Customization Mode" (top right) to open the toolbox.
- Drag & Drop extra tools (Slack, Sheets, etc.) from the library of 100+ plugins.
- Connect them to the Agent Node's special bottom ports.
Contributions are welcome! Please feel free to submit a Pull Request.
This project is open-source and available under the MIT License.