A full-stack AI-powered CRM system for managing Healthcare Professional (HCP) interactions in the pharmaceutical industry.
Features • Tech Stack • Architecture • Setup • API Docs
- 🔐 JWT Authentication — Secure login/register with role-based access control
- 👨⚕️ HCP Management — Full CRUD for Healthcare Professionals
- 📋 Interaction Logging — Track every meeting, call, and email with HCPs
- 🤖 AI Agent — LangChain + LangGraph powered agent with 6 intelligent tools
- 📊 Dashboard — Real-time activity and performance overview
- 💬 AI Chat Interface — Natural language interface for interaction management
- 📝 Follow-up Suggestions — AI-generated actionable follow-up recommendations
- 🔍 Sentiment Analysis — Automatic sentiment scoring on every interaction
Built using LangChain Tools + LangGraph State Machine:
| Tool | Description |
|---|---|
log_interaction |
Log a new HCP interaction with AI summarization |
edit_interaction |
Edit an existing interaction using natural language |
get_hcp_profile |
Retrieve complete HCP profile with full interaction history |
suggest_followup |
Generate AI-powered prioritized follow-up suggestions |
analyze_sentiment |
Analyze and score sentiment of interaction content |
search_interactions |
Search past interactions using natural language queries |
- Python 3.10+ — Core programming language
- FastAPI — High-performance REST API framework
- SQLAlchemy — ORM for database modeling and queries
- MySQL — Relational database for persistent storage
- PyJWT — JSON Web Token based authentication
- Passlib + Bcrypt — Secure password hashing
- Pydantic — Data validation and schema enforcement
- LangChain — Tool definition, LLM integration, and agent orchestration
- LangGraph — Stateful agent graph with conditional tool routing
- Groq API — Ultra-fast LLM inference (
llama-3.1-8b-instant)
- React 18 — Component-based UI library
- Vite — Lightning-fast frontend build tool
- React Router — Client-side routing and navigation
- Axios — Promise-based HTTP client for API calls
┌─────────────────────────────────────────────────┐
│ Frontend │
│ React + Vite (Port 5173) │
└───────────────────────┬─────────────────────────┘
│ HTTP / REST
┌───────────────────────▼─────────────────────────┐
│ Backend │
│ FastAPI (Port 8000) │
│ │
│ ┌──────────┐ ┌─────────────┐ ┌────────────┐ │
│ │ Auth │ │ HCP │ │Interactions│ │
│ │ API │ │ API │ │ API │ │
│ └──────────┘ └─────────────┘ └────────────┘ │
│ │
│ ┌───────────────────────────────────────────┐ │
│ │ AI Agent API │ │
│ │ LangGraph State Machine │ │
│ │ ┌─────────────────────────────────┐ │ │
│ │ │ LangChain Tools │ │ │
│ │ │ log · edit · profile · search │ │ │
│ │ │ suggest_followup · sentiment │ │ │
│ │ └────────────────┬────────────────┘ │ │
│ │ │ │ │
│ │ Groq LLM (llama-3.1-8b) │ │
│ └───────────────────────────────────────────┘ │
│ │
│ ┌───────────────────────────────────────────┐ │
│ │ SQLAlchemy ORM + MySQL Database │ │
│ │ users · hcps · interactions · followups │ │
│ └───────────────────────────────────────────┘ │
└─────────────────────────────────────────────────┘
medsync-ai/
├── backend/
│ ├── app/
│ │ ├── agent/
│ │ │ ├── graph.py # LangGraph state machine
│ │ │ └── tools/
│ │ │ ├── log_interaction.py
│ │ │ ├── edit_interaction.py
│ │ │ ├── get_hcp_profile.py
│ │ │ ├── suggest_followup.py
│ │ │ ├── analyze_sentiment.py
│ │ │ └── search_interactions.py
│ │ ├── api/
│ │ │ ├── auth.py
│ │ │ ├── hcp.py
│ │ │ ├── interactions.py
│ │ │ └── agent.py
│ │ ├── core/
│ │ │ ├── config.py
│ │ │ ├── database.py
│ │ │ └── security.py
│ │ ├── models/
│ │ │ ├── user.py
│ │ │ ├── hcp.py
│ │ │ ├── interaction.py
│ │ │ └── followup.py
│ │ ├── schemas/
│ │ │ ├── user.py
│ │ │ ├── hcp.py
│ │ │ └── interaction.py
│ │ └── main.py
│ ├── .env.example
│ └── requirements.txt
├── frontend/
│ ├── src/
│ │ ├── components/
│ │ ├── pages/
│ │ └── App.jsx
│ └── package.json
└── README.md
Make sure the following are installed on your machine:
- Python 3.10+
- Node.js 18+
- MySQL 8.0+
- Groq API Key — Get it free at console.groq.com
git clone https://github.com/itskie/medsync-ai.git
cd medsync-ai-- Login to MySQL
mysql -u root -p
-- Create the database
CREATE DATABASE medsync_db;
EXIT;# Navigate to the backend directory
cd backend
# Create a virtual environment
python3 -m venv venv
# Activate the virtual environment
source venv/bin/activate # Mac/Linux
# venv\Scripts\activate # Windows
# Install all required dependencies
pip install -r requirements.txt
# Create your environment file from the provided example
cp .env.example .env# Database Configuration
DB_HOST=localhost
DB_PORT=3306
DB_USER=root
DB_PASSWORD=your_mysql_password
DB_NAME=medsync_db
# JWT Configuration
SECRET_KEY=your_secret_key_here
ALGORITHM=HS256
ACCESS_TOKEN_EXPIRE_MINUTES=30
# Groq AI Configuration
GROQ_API_KEY=your_groq_api_key_here
# Application Settings
APP_NAME=MedSync AI
DEBUG=Truepython -m uvicorn app.main:app --reloadBackend runs at: http://127.0.0.1:8000
# Open a new terminal and navigate to the frontend directory
cd frontend
# Install all dependencies
npm install
# Start the development server
npm run devFrontend runs at: http://localhost:5173
Once the backend is running, visit the interactive Swagger UI at:
http://127.0.0.1:8000/docs
| Method | Endpoint | Description | Auth Required |
|---|---|---|---|
| POST | /auth/register |
Register a new user | ❌ |
| POST | /auth/login |
Login and receive JWT token | ❌ |
| GET | /hcps/ |
Get all HCPs | ✅ |
| POST | /hcps/ |
Create a new HCP | ✅ |
| GET | /hcps/{id} |
Get HCP by ID | ✅ |
| PUT | /hcps/{id} |
Update HCP details | ✅ |
| DELETE | /hcps/{id} |
Delete an HCP | ✅ |
| GET | /interactions/ |
Get all interactions | ✅ |
| POST | /interactions/ |
Log a new interaction | ✅ |
| GET | /interactions/{id} |
Get interaction by ID | ✅ |
| PUT | /interactions/{id} |
Update an interaction | ✅ |
| DELETE | /interactions/{id} |
Delete an interaction | ✅ |
| POST | /agent/chat |
Chat with the AI Agent | ✅ |
| GET | /agent/tools |
List all available AI tools | ✅ |
| GET | /health |
Server health check | ❌ |
This project uses JWT Bearer Token authentication.
- Register or Login to receive an
access_token - Include the token in all protected requests:
Authorization: Bearer <your_token>
POST /auth/register
{
"name": "John Doe",
"email": "john@medsync.com",
"password": "SecurePass@123",
"role": "sales_rep"
}POST /hcps/
{
"name": "Dr. Jane Smith",
"specialization": "Oncologist",
"hospital": "City Medical Center",
"city": "Mumbai",
"email": "jane.smith@citymed.com",
"phone": "9800000000"
}POST /interactions/
{
"hcp_id": 1,
"interaction_type": "meeting",
"date": "2026-03-11",
"topics_discussed": "Phase III clinical trial results and efficacy data",
"sentiment": "positive",
"outcomes": "Doctor expressed interest in prescribing the product"
}POST /agent/chat
{
"message": "Use suggest_followup tool with interaction_id 1",
"interaction_id": 1
}| Variable | Description | Example |
|---|---|---|
DB_HOST |
MySQL host address | localhost |
DB_PORT |
MySQL port number | 3306 |
DB_USER |
MySQL username | root |
DB_PASSWORD |
MySQL password | yourpassword |
DB_NAME |
Database name | medsync_db |
SECRET_KEY |
JWT signing secret key | random_secure_string |
ALGORITHM |
JWT hashing algorithm | HS256 |
ACCESS_TOKEN_EXPIRE_MINUTES |
Token expiry time in minutes | 30 |
GROQ_API_KEY |
Groq LLM API key | gsk_... |
DEBUG |
Enable debug mode | True / False |
For production deployment:
- Set
DEBUG=Falsein environment variables - Use a strong randomly generated
SECRET_KEY - Store all secrets using environment variables — never hardcode them
- Use AWS RDS or PlanetScale for managed MySQL
- Deploy backend on Railway, Render, or AWS EC2
- Deploy frontend on Vercel or Netlify
Shobhit Kumar Singh
- GitHub: @itskie
- LinkedIn: linkedin.com/in/itskie
This project is licensed under the MIT License.
Built with ❤️ by Shobhit Singh | MedSync AI © 2026