NetsolChatbot

An intelligent, multi-tool AI assistant built for NetsolTech — powered by LangGraph, Google Gemini, and React.
Supports RAG document Q&A, live web search, Google Calendar, natural language SQL queries, and interactive data visualizations.

---

📸 Screenshots

📈 Natural language → SQL → Line Chart	📊 Revenue by City — Bar Chart
🔀 Multi-tool in one query — Calendar events + Top 5 customers chart

---

✨ Features

Feature	Description
🧠 LangGraph Agent	Multi-step tool-calling agent with streaming responses
📄 RAG (Document Q&A)	Upload PDF, DOCX, or TXT — ask questions from content
📊 Text-to-SQL + Charts	Natural language → SQL → interactive bar, line, pie charts
🌐 Web Search	Real-time search via Tavily API
📅 Google Calendar	Check and create meetings via service account
💬 Chat History	Persistent multi-thread conversations with sidebar
⚡ Streaming	Real-time token-by-token response streaming

---

🏗️ Architecture

User (React Frontend)
        │
        ▼
  FastAPI Backend
        │
        ▼
  LangGraph Graph
   ┌────┴────┐
   ▼         ▼
Retriever   LLM Node (Gemini)
(ChromaDB)      │
                ├── Tool: query_business_database → SQLite
                ├── Tool: web_search → Tavily API
                ├── Tool: get_upcoming_events → Google Calendar
                └── Tool: create_calendar_event → Google Calendar

Request Flow:

1. User message → FastAPI /api/chat
2. LangGraph retriever_node → similarity search in ChromaDB (top 10 chunks)
3. LangGraph llm_node → Gemini decides: answer from context OR call a tool
4. If tool called → tools_node executes → result back to LLM
5. Final response streamed chunk-by-chunk to frontend
6. Frontend parses <chart_data> tags → renders Recharts visualizations

---

🧰 Tech Stack

Layer	Technology
Backend	Python 3.10+, FastAPI, Uvicorn
AI / Agents	LangGraph, LangChain, Google Gemini 2.5 Flash-Lite
Embeddings	Sentence-Transformers (all-MiniLM-L6-v2)
Vector Store	ChromaDB
Database	SQLite (chat history + business data)
Frontend	React 18, Vite, Tailwind CSS
Charts	Recharts
Icons	Lucide React
Search	Tavily Search API
Calendar	Google Calendar API (Service Account)

---

📁 Project Structure

NetsolChatbot/
├── backend/
│   ├── main.py                    # FastAPI app + endpoints
│   ├── run.py                     # Uvicorn entry point
│   ├── src/
│   │   ├── graph.py               # LangGraph workflow
│   │   ├── state.py               # AgentState definition
│   │   ├── nodes/
│   │   │   ├── retriever.py       # ChromaDB similarity search
│   │   │   └── llm_responder.py   # Gemini LLM node
│   │   ├── rag/
│   │   │   ├── embeddings.py      # SentenceTransformer wrapper
│   │   │   ├── vector_store.py    # ChromaDB setup
│   │   │   └── document_loader.py # PDF/DOCX/TXT processing
│   │   ├── tools/
│   │   │   ├── calendar_tool.py   # Google Calendar tools
│   │   │   ├── search_tool.py     # Tavily web search
│   │   │   └── sql_tool.py        # Text-to-SQL tool
│   │   ├── models/
│   │   │   └── llm_factory.py     # Gemini client factory
│   │   ├── database.py            # SQLite chat history
│   │   └── utils/
│   │       └── helpers.py         # Title generation etc.
│   ├── scripts/
│   │   └── create_sample_db.py    # Sample business data seeder
│   ├── credentials/               # Google service account (gitignored)
│   ├── data/                      # SQLite DBs + ChromaDB (gitignored)
│   ├── .env                       # Environment variables (gitignored)
│   └── requirements.txt
├── frontend/
│   ├── src/
│   │   ├── components/
│   │   │   ├── chat/
│   │   │   │   ├── ChatWindow.jsx
│   │   │   │   ├── MessageList.jsx    # Chart tag parser + renderer
│   │   │   │   ├── MessageInput.jsx   # File upload + message send
│   │   │   │   ├── ChartMessage.jsx   # Recharts visualization
│   │   │   │   └── TypingIndicator.jsx
│   │   │   └── layout/
│   │   │       ├── ChatLayout.jsx
│   │   │       └── Sidebar.jsx
│   │   ├── context/
│   │   │   └── ChatContext.jsx        # Global state + streaming logic
│   │   └── services/
│   │       └── api.js                 # Backend API calls
│   └── package.json
├── docs/
│   └── screenshots/               # README screenshots
├── .gitignore
└── README.md

---

⚙️ Setup

Prerequisites

• Python 3.10+
• Node.js 18+
• A Google Cloud project with:
  • Gemini API enabled → Get API key
  • Google Calendar API enabled → Service Account setup
• Tavily API key → Get free key (1000 searches/month free)

---

1. Clone

git clone https://github.com/tashfeen786/NetsolChatbot.git
cd NetsolChatbot

---

2. Backend Setup

cd backend
python -m venv venv

# Windows
.\venv\Scripts\activate
# Linux/Mac
source venv/bin/activate

pip install -r requirements.txt

Create backend/.env:

# Gemini
GOOGLE_API_KEY=your_gemini_api_key
MODEL_NAME=gemini-2.5-flash-lite

# Google Calendar (Service Account)
GOOGLE_SERVICE_ACCOUNT_FILE=credentials/your-service-account.json
GOOGLE_CALENDAR_ID=your_calendar_id@group.calendar.google.com

# Tavily Web Search
TAVILY_API_KEY=tvly-your_tavily_key

# Business Database
BUSINESS_DB_PATH=data/business.db

Place your Google Service Account JSON in backend/credentials/.

Seed the sample business database:

python scripts/create_sample_db.py

Start the backend:

python run.py
# API running at http://localhost:8000
# Docs at http://localhost:8000/docs

---

3. Frontend Setup

cd ../frontend
npm install
npm run dev
# App running at http://localhost:5173

---

🌐 API Reference

Method	Endpoint	Description
GET	/api/threads	List all chat threads
POST	/api/threads	Create a new thread
GET	/api/threads/{id}/messages	Get messages for a thread
POST	/api/chat	Send message (streaming response)
POST	/api/upload	Upload PDF / DOCX / TXT file

---

💡 Usage Examples

Document Q&A

Click 📎 → upload a PDF → ask:
"Summarize the key points from the uploaded document"

Business Data + Chart

"Show total revenue by city as a bar chart"
"Which account manager has the highest sales?"
"Show monthly revenue trend for the last 6 months as a line chart"

Calendar

"What meetings do I have today?"
"Schedule a team sync tomorrow at 3 PM to 4 PM"

Web Search

"What is the current USD to PKR exchange rate?"
"Search for latest news about NetsolTech"

Multi-tool (one query, two tools)

"What meetings do I have today and show top 5 customers as a chart"

---

🔐 Security Notes

• .env files and credentials/ are excluded from version control via .gitignore
• SQL tool only allows SELECT queries — DROP, DELETE, INSERT, UPDATE and other destructive keywords are blocked
• File uploads are validated by extension and size (max 10MB)
• API keys are never exposed in source code or frontend bundles

---

🛠️ Troubleshooting

Issue	Fix
Server crashes silently on first request	Ensure os.environ["KMP_DUPLICATE_LIB_OK"] = "TRUE" is the first line in run.py
503 UNAVAILABLE from Gemini	API rate limit — wait a few seconds and retry
Calendar returns no events	Verify service account email is shared on the calendar with "Make changes to events" permission
Chart not rendering	Check browser console for <chart_data> parse errors; ensure recharts is installed
File upload fails	Check file is PDF/DOCX/TXT and under 10MB

---

🚀 Deployment

Backend — Railway or Render

• Set all .env variables in the platform dashboard
• Set start command: python run.py

Frontend — Vercel or Netlify

• Set VITE_API_URL=https://your-backend-url.com
• Build command: npm run build

---

🤝 Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature/your-feature
3. Commit: git commit -m "Add your feature"
4. Push: git push origin feature/your-feature
5. Open a Pull Request

---

📄 License

This project is licensed under the MIT License.

---

👤 Author

Tashfeen
GitHub: @tashfeen786
Project: NetsolChatbot

---

Built with ❤️ using LangGraph · Gemini · React