Deep Thinker

A full-stack AI research agent that breaks complex questions into parallel research threads, searches the web, summarizes findings, and synthesizes a final report - all streamed live to the UI.

Screenshots

Running - parallel researcher threads searching the web, results streaming in real time

Complete - synthesized markdown report with all sources

Architecture

frontend (Next.js 16)  ←→  backend (FastAPI + LangGraph)
                              ↓
              START → planner → [Send ×N] → researcher (parallel)
                                                  ↓ (merge)
                                              summarizer
                                                  ↓
                                              synthesizer → END

Monorepo

Directory	Stack
frontend/	Next.js 16 + React 19 + TypeScript + Tailwind CSS v4
backend/	FastAPI + LangGraph + langchain-openai + Tavily

Quick Start

Backend

cd backend
uv sync
cp .env.example .env   # fill in OPENAI_API_KEY, TAVILY_API_KEY
uv run main.py

Frontend

cd frontend
pnpm install
cp .env.example .env.local   # set NEXT_PUBLIC_API_URL
pnpm dev

App runs at http://localhost:3000, backend at http://localhost:8000.

Environment Variables

Backend (backend/.env)

OPENAI_API_KEY=sk-...
GROQ_API_KEY=gsk_...       # optional, for Groq provider
TAVILY_API_KEY=tvly-...
LLM_PROVIDER=openai
LLM_MODEL=gpt-4o-mini

Frontend (frontend/.env.local)

NEXT_PUBLIC_API_URL=http://localhost:8000