Multi-Agent Researcher

---

🚀 Overview

Multi-Agent Researcher is a full-stack, AI-powered research assistant platform. It automates the process of researching, synthesizing, and reporting on complex topics using a pipeline of specialized AI agents. The system features real-time progress tracking, authentication, and a modern, user-friendly interface.

---

🎬 Demo

Click an image below to play the full demo GIF!

1. Submitting a Research Query


2. Viewing Research Results

---

Contents

• Overview
• Demo
• Features
• Architecture & Workflow
  • How it Works
  • Agent Pipeline
• Tech Stack Explained
• Setup (Local Development)
  • Backend
  • Frontend
• API Endpoints
  • GET /api/health
  • POST /api/research-agent
• User Experience Flow
• Deployment
• Contributing
• License

---

✨ Features

• Automated Multi-Agent Research: Submit a topic and let a pipeline of AI agents handle web search, outlining, research, review, synthesis, and publishing.
• Real-Time Progress Tracking: See each agent's status and progress live as your research is processed.
• Authentication & User History: Secure login, personalized settings, and a dashboard of all your past research queries and results.
• Rich Results & Export: Download research reports, view sources and perspectives, and export or share results in multiple formats.
• Modern, Responsive UI: Beautiful, accessible interface that works on desktop and mobile.
• Customizable Workflow: Easily add, remove, or modify agents and research steps.
• Supabase Integration: Real-time database and authentication for seamless user experience.
• Gemini AI & LangGraph: Advanced content generation and multi-agent orchestration.

---

🧩 Architecture & Workflow

How it Works

• User interacts with the web UI to submit research queries and view results.
• Frontend authenticates users, submits queries, and displays real-time progress/results.
• Supabase manages user accounts, stores research data, and streams updates to the UI.
• Backend receives research requests, authenticates via Supabase JWT, and launches a multi-agent workflow.
• LangGraph Agents (in the backend) perform research, synthesis, review, and reporting, updating progress in the database.

Agent Pipeline

Each agent is a specialized AI module:

• Browser Agent: Performs initial web research and gathers sources.
• Editor Agent: Creates a structured outline for the research topic.
• Researcher Agent: Conducts in-depth research and drafts content for each section.
• Reviewer Agent: Fact-checks and revises the draft for accuracy and clarity.
• Writer Agent: Synthesizes the final report in the requested format.
• Publisher Agent: Saves the results and updates the query status.

---

🛠️ Tech Stack Explained

• Backend:
  • FastAPI: High-performance Python web framework for building APIs.
  • LangGraph: Orchestrates a pipeline of AI agents, each handling a stage of the research process.
  • Supabase: Provides authentication, real-time database, and secure data storage.
  • Gemini AI: Used for advanced content generation and analysis.
• Frontend:
  • React + Vite: Fast, modern UI with instant feedback and hot reloading.
  • Supabase JS: Handles authentication and real-time data updates in the browser.
  • Tailwind CSS & shadcn/ui: For a beautiful, responsive, and accessible design.

---

Setup (Local Development)

Backend

1. Create and activate a Python virtual environment (recommended):

   cd backend
   python3 -m venv .venv
   source .venv/bin/activate  # On Windows use: .venv\Scripts\activate

2. Install dependencies:

   pip install -r requirements.txt

3. Configure environment variables:

   • Create a file named .env in the backend directory (if it doesn't exist) and add the following:

     SUPABASE_URL=<your-supabase-url>
     SUPABASE_ANON_KEY=<your-supabase-anon-key>
     GEMINI_API_KEY=<your-gemini-api-key>
     FRONTEND_URL=<your-local-frontend-url>
     PORT=8000

   • Replace the placeholders with your actual values.

4. Run the backend server:

   uvicorn main:app --reload --host 0.0.0.0 --port 8000

Frontend

1. Install dependencies:

   npm install

2. Configure environment variables:

   • Create a file named .env in the root directory (if it doesn't exist) and add the following:

     VITE_SUPABASE_URL=<your-supabase-url>
     VITE_SUPABASE_PUBLISHABLE_KEY=<your-supabase-anon-key>
     VITE_BACKEND_URL=http://localhost:8000

   • Replace the placeholders with your actual values.

3. Run the frontend dev server:

   npm run dev

---

📄 API Endpoints

• GET /api/health — Health check (returns status of backend)
• POST /api/research-agent — Start a research workflow (requires JWT)
  • Body:

  { "queryId": "uuid-of-research-query" }

All endpoints are documented and protected as needed. The main workflow endpoint requires a valid Supabase JWT for user context and Row-Level Security.

---

🧑‍💻 User Experience Flow

1. Sign up or log in (Supabase authentication)
2. Submit a research query (topic, depth, perspectives, sources, etc.)
3. Track progress in real-time as each agent completes its task
4. View results: Download, review sources, perspectives, and provide feedback
5. Access history: All your past queries and results are available in your dashboard

---

📦 Deployment

• Backend: Deploy on Railway, Render, or any platform supporting FastAPI + Python. Set environment variables for Supabase, Gemini and deployed frontend URL.

• Frontend: Deploy on Vercel, Netlify, or any static hosting. Set environment variables for Supabase and deployed backend URL.

---

🤝 Contributing

We welcome contributions from the community! To contribute:

1. Fork this repository.
2. Create a new branch for your feature or bugfix.
3. Make your changes and commit them with clear messages.
4. Open a Pull Request describing your changes.

Thank you for helping improve Multi-Agent Researcher!

---

📚 License

This project is licensed under the MIT License.