🧬 Pramana.ai

Master biomedical evidence. Without the silos.

AI-powered biomedical research platform built with LangGraph, Groq, and React 19

Quick Start • Features • Agents • Architecture • Deployment • Documentation

---

✨ What Makes This Special

Pramana.ai unifies biomedical research workflows into a single AI-powered platform:

• 🧬 Purpose-Built for Biomedical Research — Drug discovery, clinical trials, market intelligence
• 🤖 5 Specialized AI Agents — Portfolio Strategist, Deep Researcher, Chat Assistant, Math Solver, MCP Agent
• ⚡ Lightning-Fast Inference — Powered by Groq's LLaMA 3.3 70B models
• 🔌 Extensible Tool Ecosystem — Model Context Protocol (MCP) integration
• 📊 Real-Time Streaming — Watch AI agents think and work live
• 🧩 Workflow Canvas — Visualize multi-agent orchestration as a node graph. Run demo queries to animate each step (pending → processing → completed) and click nodes to inspect tasks and data sources.
• 🎨 Beautiful Neo-Brutalist UI — Modern React 19 + Framer Motion design
• 🔐 Production-Ready Auth — JWT authentication, SQLite/PostgreSQL persistence

---

🚀 Quick Start

Prerequisites

• Node.js 18+ & npm
• Python 3.11+ & UV package manager
• API Keys: Groq (required), SerpAPI (for Deep Researcher)

Installation

# Clone and enter directory
git clone https://github.com/darved2305/ggw_eytechathon.git
cd ggw_eytechathon

# Configure environment
cd backend
cp .env.example .env
# Edit .env with your GROQ_API_KEY and SERPAPI_API_KEY

# Install dependencies
uv sync                                    # Backend
cd ../frontend && npm install              # Frontend

# Start all servers
cd .. && make dev

Access the app:

• Landing Page: http://localhost:5173
• Main App: http://localhost:5173/app
• Workflow Canvas: http://localhost:5173/workflow
• Chat Interface (legacy): http://localhost:5173/chat
• LangGraph API: http://localhost:2024/docs

💡 New to the project? Check out the Development Guide for detailed setup instructions.

---

🤖 AI Agents

Pramana.ai includes five specialized AI agents, each optimized for different research workflows:

💼 Portfolio Strategist

AI-powered pharmaceutical innovation discovery and opportunity scoring

Analyzes drug opportunities through 6 parallel data sources (IQVIA, Trials, Patents, EXIM, Internal, Web), applies 7 decision heuristics, and generates evidence-backed opportunity scores.

Key Signals:

• HIGH_WHITESPACE — Strong unmet need with few active trials
• PATENT_WINDOW_OPEN — Key patents expiring soon
• FRAGMENTED_MARKET — Low concentration, entry opportunity

📖 View detailed agent documentation

🔍 Deep Researcher

Advanced web research with iterative query refinement

Generates optimized search queries, performs parallel web research via SerpAPI, reflects on findings, and produces citation-first reports.

Configuration: Supports custom search query counts, max loops, and model selection.

📖 View detailed agent documentation

💬 Chat Assistant

General-purpose conversational AI with context awareness

Natural language conversations powered by Groq's fast inference. Best for quick Q&A and general assistance.

🧮 Math Solver

Safe mathematical expression evaluation

Supports arithmetic operations, mathematical functions (sqrt, sin, cos, log, exp), and constants (pi, e).

🔧 MCP Agent

Model Context Protocol integration for external tools

Connects to MCP servers (Filesystem, Brave Search) for extensible tool capabilities.

📖 Full agent documentation →

---

🏗️ Architecture

Tech Stack

Frontend: React 19, TypeScript, Tailwind CSS v4, Framer Motion, Vite
Backend: Python 3.11+, LangGraph, LangChain, Groq, FastAPI
Infrastructure: Redis (streaming), PostgreSQL (persistence), Docker

┌─────────────────────────────────────────────────────────┐
│                    React 19 Frontend                     │
│        (Neo-brutalist UI + Real-time Streaming)          │
└──────────────────────┬──────────────────────────────────┘
                       │ HTTP/SSE
┌──────────────────────┴──────────────────────────────────┐
│               LangGraph Backend (Port 2024)              │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌─────────┐ │
│  │Portfolio │  │  Deep    │  │   Chat   │  │   MCP   │ │
│  │Strategist│  │Researcher│  │ Assistant│  │  Agent  │ │
│  └──────────┘  └──────────┘  └──────────┘  └─────────┘ │
└──────────────────────┬──────────────────────────────────┘
                       │
┌──────────────────────┴──────────────────────────────────┐
│              FastAPI Auth Server (Port 8000)             │
│         JWT Authentication + SQLAlchemy ORM              │
└──────────────────────┬──────────────────────────────────┘
                       │
┌──────────────────────┴──────────────────────────────────┐
│     Redis (Streaming)     │    PostgreSQL (State)       │
└───────────────────────────┴─────────────────────────────┘

📖 Detailed architecture documentation →

---

🚢 Deployment

Development (2 servers + optional Auth API)

# Option 1: Use Makefile (recommended)
make dev

# Option 2: Manual startup
cd backend && uv run langgraph dev --no-browser        # Terminal 1 (port 2024)
cd frontend && npm run dev                             # Terminal 2 (port 5173)

# Optional: Auth API (needed for /login, /register, and persisted chat history)
cd backend && uv run uvicorn src.auth.app:app --reload --port 8000

# If uvicorn isn't available in your env:
#   cd backend && uv add uvicorn

Docker (Production)

docker build -t pramana-ai -f Dockerfile .
GROQ_API_KEY=xxx SERPAPI_API_KEY=xxx docker-compose up

📖 Full deployment guide → | Configuration reference →

---

📚 Documentation

Document	Description
Agents	Detailed agent behavior, heuristics, and configuration
Architecture	Full stack, infrastructure, and LangGraph internals
API Reference	LangGraph endpoints and backend API routes
Configuration	Environment variables, models, and MCP servers
Deployment	Production checklist, Docker, and scaling
Development	Testing, linting, and contribution workflow
Frontend	UI components, landing page, and design system
Backend	Authentication, database schema, and persistence
Troubleshooting	Common errors and fixes

---

🤝 Contributing

We welcome contributions! Please read our Development Guide for setup instructions and coding standards.

Quick contribution workflow:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push and open a Pull Request

Ideas for contributions: New agents, data connectors, UI improvements, documentation, tests

---

📄 License

This project is licensed under the Apache License 2.0. See LICENSE for details.

---