Deep Agent Chat

An intelligent multi-route AI assistant that combines:

• Standard conversational chat
• Retrieval-Augmented Generation (RAG)
• Real-time web research
• Weather lookup
• Session memory
• MCP tool integration

The system automatically decides whether a user query should be handled by:

1. Chat Node → normal conversation
2. RAG Node → answer using uploaded documents
3. Research Node → live internet search and weather tools

---

Features

Smart Query Routing

The assistant automatically routes each prompt:

Query Type	Route
General chat	Chat Node
Document-related	RAG Node
Real-time info	Research Node

Examples

• Explain machine learning → Chat
• Summarize my uploaded PDF → RAG
• Weather in New York → Research
• Latest AI news → Research

---

RAG Support

Uses PostgreSQL + pgvector for:

• Semantic search
• PDF retrieval
• Context-aware answers
• Document understanding

---

Real-Time Research

Supports:

• Live web search via Tavily
• Weather lookup
• Multi-step planning
• Final answer synthesis

---

Session Memory

Maintains short-term memory per session:

• Stores previous messages
• Preserves conversation context
• Trims old messages automatically

---

Project Structure

Deep_Agent_Chat/
│── auth.py
│── database.py
│── embedding.py
│── models.py
│── rag_pgvector.py
│── title_app.py
│── mcp_weather_tool.py
│── weather_mcp1.py
│── tavily_mcp_server.py

---

```bash
Installation 
1. Clone Repository
git clone https://github.com/yourusername/deep-agent-chat.git
cd deep-agent-chat

2. Create Virtual Environment
python -m venv venv

Linux / macOS
source venv/bin/activate

Windows
venv\Scripts\activate

3. Install Dependencies
pip install -r requirements.txt

4. Create .env File
GROQ_API_KEY=your_groq_api_key
DATABASE_URL=postgresql://user:password@localhost:5432/deep_agent

Required Services
PostgreSQL + pgvector

Enable extension:

CREATE EXTENSION vector;


Tavily MCP Server

Runs on:

http://localhost:8002/mcp

Weather MCP Server

Runs on:

http://localhost:8001/mcp

---

Technologies Used:

Python
LangChain
Groq LLM
PostgreSQL
pgvector
MCP Protocol
AsyncIO

---

Future Improvements

Planned enhancements:

Streaming responses
Multi-document retrieval
Long-term memory
Source citations
Better tool routing
Analytics dashboard

---

License

MIT License

---

Contributing

Contributions are welcome.

Steps
Fork the repository
Create your feature branch
Commit your changes
Push to the branch
Open a pull request