🚀 AI Workflow Automation & Multi-Agent Research Platform

A lightweight, high-performance platform for constructing automated AI workflows and generating comprehensive research reports using a multi-agent orchestration pipeline.

---

✨ Overview

This platform combines:

• 🤖 Multi-Agent Research System
• ⚡ Workflow Automation Engine
• 📚 Local RAG (Retrieval-Augmented Generation)
• 🔍 Semantic Search
• 🔐 Authentication & Authorization
• 📁 File Upload & Knowledge Management
• 📊 Real-Time Execution Tracking

The project is designed to run with minimal setup while providing enterprise-style workflow orchestration capabilities.

---

🏗️ Tech Stack

Backend

• FastAPI
• SQLAlchemy ORM
• SQLite / PostgreSQL
• Google Gemini AI
• JWT Authentication
• Server-Sent Events (SSE)

Frontend

• React SPA
• Tailwind CSS v4
• Zero-build architecture
• Browser-side JSX transpilation via Service Worker

AI & Search

• Multi-Agent Pipeline
• RAG Architecture
• Vector Similarity Search
• Embedding Generation

---

📂 Project Structure

ai-orchestrator/
│
├── backend/
│   ├── main.py
│   ├── models.py
│   ├── database.py
│   ├── auth.py
│   ├── agents.py
│   ├── vector_store.py
│   ├── workflow_engine.py
│   ├── requirements.txt
│   └── routes/
│       ├── auth.py
│       ├── agents.py
│       ├── workflows.py
│       ├── search.py
│       └── files.py
│
├── frontend/
│   ├── index.html
│   ├── sw.js
│   ├── app.js
│   └── components/
│       ├── auth/
│       ├── workflow/
│       ├── research/
│       ├── search/
│       └── files/
│
├── .gitignore
├── .env.example
└── README.md

---

⚙️ Environment Configuration

1. Clone Repository

git clone https://github.com/<your-username>/ai-orchestrator.git
cd ai-orchestrator

---

2. Configure Environment Variables

Copy the environment template:

cp .env.example backend/.env

Open backend/.env and add:

GEMINI_API_KEY=your_api_key_here

Get Gemini API Key

Visit:

https://aistudio.google.com/

Generate a new API key and paste it into the .env file.

---

Optional: Use PostgreSQL / Supabase

Instead of SQLite:

DATABASE_URL=postgresql://postgres:<password>@<host>.supabase.co:5432/postgres

SUPABASE_URL=https://<project-ref>.supabase.co

SUPABASE_SERVICE_ROLE_KEY=<service-role-key>

---

Enable PGVector

Run inside Supabase SQL Editor:

CREATE EXTENSION IF NOT EXISTS vector;

---

🔒 Security Notes

Never commit:

.env
database credentials
API keys
service role keys
SQLite database files

The provided .gitignore already excludes these files.

---

🖥️ Backend Setup

Navigate to backend:

cd backend

Create Virtual Environment

Windows

python -m venv venv

venv\Scripts\activate

Linux / macOS

python -m venv venv

source venv/bin/activate

---

Install Dependencies

pip install -r requirements.txt

---

Run Server

python main.py

Backend runs at:

http://localhost:8000

The SQLite database will be created automatically.

---

🎨 Frontend Setup

Because the frontend uses a Service Worker for browser-side JSX compilation, it must be served through an HTTP server.

---

Option A (Recommended)

Run backend:

python main.py

Open:

http://localhost:8000

FastAPI automatically serves frontend assets.

---

Option B (Standalone Frontend)

cd frontend

python -m http.server 3000

Open:

http://localhost:3000

Ensure backend is running on port 8000.

---

🤖 Multi-Agent Research Pipeline

The Research module uses a chain of specialized AI agents:

1. Planner Agent

Creates structured research sections.

2. Research Agent

Collects information and notes.

3. Critic Agent

Reviews findings for:

• Bias
• Logic gaps
• Missing information

4. Writer Agent

Generates the final report.

5. Citation Agent

Adds references and bibliography.

---

Real-Time Streaming

Progress is streamed using:

Server-Sent Events (SSE)

Users can watch research progress live.

---

⚙️ Workflow Automation Engine

Build custom DAG workflows using:

• Triggers
• LLM Nodes
• Delay Nodes
• API Request Nodes
• Conditional Logic

Variable Binding

{{node_id.output}}

Outputs can be passed between nodes dynamically.

---

📚 Local RAG (Retrieval-Augmented Generation)

Upload your documents through the Files Manager.

The system will:

1. Chunk documents
2. Generate embeddings
3. Store vectors
4. Enable semantic retrieval

Supported use cases:

• Knowledge bases
• Internal documentation
• Research archives
• Company policies

---

🔍 Semantic Search

Search across uploaded knowledge using vector similarity.

Features:

• Context-aware retrieval
• Similarity ranking
• Fast local lookup
• AI-enhanced responses

---

🔐 Authentication

Supports:

• JWT Authentication
• Password Hashing (bcrypt)
• Protected API Routes
• Session Management

---

📈 Future Enhancements

Planned features:

• Role-Based Access Control (RBAC)
• WebSockets
• Redis Caching
• Celery Workers
• Agent Marketplace
• Kubernetes Deployment
• LangGraph Integration
• OpenTelemetry Observability
• Multi-Tenant Architecture

---

🤝 Contributing

1. Fork the repository
2. Create a feature branch

git checkout -b feature/new-feature

3. Commit changes

git commit -m "Add new feature"

4. Push

git push origin feature/new-feature

5. Open a Pull Request

---

📜 License

MIT License

Feel free to use, modify, and distribute.

---

⭐ Support

If you found this project useful:

⭐ Star the repository

🍴 Fork the project

🛠️ Contribute improvements

📢 Share it with others