AI Researcher

An autonomous research assistant that decomposes complex queries into hypotheses, launches agents to investigate them using Deepseek Reasoner, and synthesizes a final report.

Architecture

The system consists of two main parts:

1. Backend (FastAPI):

   • Orchestrator: Breaks down the user's query into sub-tasks (hypotheses).
   • Agents: Autonomous workers that "reason" about each hypothesis using the Deepseek API.
   • Deepseek Integration: Uses the deepseek-reasoner (or compatible) model for high-level logic.
   • PDF Service: Generates downloadable reports.

2. Frontend (React + Vite):

   • Live Dashboard: Visualizes the agents' thinking process in real-time.
   • Technical Aesthetic: Designed with a minimal, dark-mode "terminal" look.

Quick Start

Prerequisites

• Python 3.8+
• Node.js 16+
• Deepseek API Key

1. Start the Backend

Navigate to the backend folder:

cd backend
pip install -r requirements.txt
# Create .env file with DEEPSEEK_API_KEY=...
python -m uvicorn main:app --reload --host 0.0.0.0 --port 8000

2. Start the Frontend

Navigate to the frontend folder:

cd frontend
npm install
npm run dev

3. Use the App

Open http://localhost:5173 in your browser. Enter a research query (e.g., "The future of solid-state batteries") and watch the agents work.

Features

• Deepseek Reasoner: Powered by advanced reasoning models.
• Live Streaming: See the AI's "thought process" as it happens.
• PDF Export: Download a professionally formatted report of the findings.