🤖 MultiGenAI

An Intelligent Multi-Domain Generative AI Workspace

Status: ✅ Completed & Tested

MultiGenAI is a personalized, general-purpose Generative AI assistant that understands what the user wants automatically and generates context-aware responses.

Instead of forcing users to manually select a domain, MultiGenAI uses an LLM-powered Smart Intent Detection system to understand the user's request and dynamically adapt the response.

---

🚀 Live Demo

🔗 Launch MultiGenAI

---

✨ Features

• 🧠 Smart Intent Detection
• 🎯 Automatic domain detection
• 📌 Subject detection
• 💡 Intent detection
• 📝 Task detection
• ⚡ Urgency detection
• 🤖 Qwen LLM powered by Groq
• 💬 Context-aware conversations
• 🔐 User Registration and Login
• 🔒 Secure password hashing
• 👤 User-specific private chat history
• 💾 Persistent SQLite database
• 🔎 Chat history search
• 🆕 New Chat management
• ⚡ Streaming AI responses
• 🏥 Medical safety-aware responses
• 🌐 General-purpose AI capability
• 🎨 Clean dark-themed interface

---

🖥️ Screenshots

🔐 Login / Registration

💬 Main Chat Interface

💬 Chat Interface

🧠 Smart Mode

The main feature of MultiGenAI is its Smart Mode.

Users do not need to manually select a domain.

The system automatically analyzes the user's message and identifies:

Domain
Subject
Intent
Task
Urgency

Example

User:

Tomorrow my chemistry exam. Give me 10 MCQs with answers.

MultiGenAI understands:

Domain   → Education
Subject  → Chemistry
Intent   → Exam Preparation
Task     → Generate MCQs with answers
Urgency  → High

The detected context is then passed to the Prompt Builder to create a more relevant instruction for the main LLM.

---

🏗️ System Architecture

                    ┌───────────────────┐
                    │    User Message   │
                    └─────────┬─────────┘
                              │
                              ▼
                  ┌───────────────────────┐
                  │    Intent Detector    │
                  │                       │
                  │ • Domain              │
                  │ • Subject             │
                  │ • Intent              │
                  │ • Task                │
                  │ • Urgency             │
                  └───────────┬───────────┘
                              │
                              ▼
                  ┌───────────────────────┐
                  │    Prompt Builder     │
                  └───────────┬───────────┘
                              │
                              ▼
                  ┌───────────────────────┐
                  │ Conversation Context  │
                  └───────────┬───────────┘
                              │
                              ▼
                  ┌───────────────────────┐
                  │      Qwen + Groq      │
                  └───────────┬───────────┘
                              │
                              ▼
                  ┌───────────────────────┐
                  │    AI Response        │
                  └───────────────────────┘

---

🔐 Authentication

MultiGenAI supports real user authentication.

User
 │
 ├── Register
 │      │
 │      ▼
 │   User Account
 │
 └── Login
        │
        ▼
   Authenticated User
        │
        ▼
      Chat
        │
        ▼
      user_id
        │
        ▼
 User-specific Chat History

Each chat is associated with the authenticated user's user_id.

This ensures that one user cannot access another user's conversations through the application's database queries.

---

💾 Database

MultiGenAI uses SQLite for persistent storage.

Users Table

users
├── id
├── name
├── email
└── password_hash

Chats Table

chats
├── id
├── user_id
├── title
├── domain
└── messages

The user_id connects every conversation to its owner.

---

🧩 Project Structure

MultiGenAI/
│
├── tests/
│   ├── test_intent.py
│   └── test_prompt.py
│
├── screenshots/
├── streamlit/
│
├── app.py
├── auth.py
├── database.py
├── intent_detector.py
├── prompt_builder.py
│
├── requirements.txt
├── README.md
├── .gitignore
│
├── .env
└── multigenai.db

Important

The following files should not be committed to GitHub:

.env
multigenai.db
venv/
__pycache__/

---

📁 Module Description

File	Purpose
app.py	Main Streamlit application, UI and chat flow
auth.py	User registration, login and password verification
database.py	SQLite database initialization
intent_detector.py	Detects domain, subject, intent, task and urgency
prompt_builder.py	Creates dynamic context-aware prompts
tests/test_intent.py	Tests Smart Intent Detection
tests/test_prompt.py	Tests Prompt Builder

---

🛠️ Technology Stack

Programming Language

• Python

Generative AI

• Qwen
• Groq API
• Large Language Models
• Prompt Engineering
• Intent Detection

Framework

• Streamlit

Database

• SQLite

Authentication

• Werkzeug Password Hashing

Python Libraries

• Streamlit
• Groq
• python-dotenv
• Werkzeug

---

🎯 Use Cases

MultiGenAI is designed as a general-purpose AI workspace.

💻 Programming

Example:

Why does my Python code give a ValueError?

MultiGenAI identifies the request as a programming and debugging task.

---

🤖 AI & Machine Learning

Example:

Explain overfitting with a real-world example.

The system identifies the topic and provides a practical explanation.

---

📚 Education

Example:

Tomorrow is my chemistry exam.
Give me 20 MCQs with answers.

The system understands the educational context and exam preparation requirement.

---

💼 Career

Example:

Analyze this Data Science internship job description
and tell me what skills I need.

The system identifies the career-related task and provides relevant guidance.

---

✍️ Content Writing

Example:

Write a professional LinkedIn post about my new AI project.

The system adapts the response according to the requested writing task.

---

🌎 General Questions

Example:

Why is the sky blue?

MultiGenAI can also handle questions outside predefined domains.

---

🏥 Medical Safety

For health-related questions, MultiGenAI follows safety-aware response behavior.

The system is designed to:

• Provide general educational information
• Avoid claiming to diagnose users
• Avoid prescribing personalized medication or dosage
• Recommend appropriate urgent medical care when symptoms may indicate an emergency

MultiGenAI is an AI assistant and does not replace a qualified medical professional.

---

🔄 Complete Workflow

User
 ↓
Authentication
 ↓
Enter Message
 ↓
Smart Intent Detection
 ↓
Domain + Subject + Intent + Task + Urgency
 ↓
Prompt Builder
 ↓
Conversation Context
 ↓
Qwen Model via Groq
 ↓
Streaming Response
 ↓
Save Conversation
 ↓
SQLite Database

---

🧪 Testing

MultiGenAI includes separate testing scripts for the Smart Mode components.

Test Intent Detection

python tests/test_intent.py

Test Prompt Builder

python tests/test_prompt.py

These tests help verify the individual components before using them inside the main application.

---

⚙️ Installation

1. Clone the Repository

git clone <YOUR_GITHUB_REPOSITORY_URL>

Then enter the project directory:

cd MultiGenAI

---

2. Create Virtual Environment

python -m venv venv

Windows

venv\Scripts\activate

---

3. Install Dependencies

pip install -r requirements.txt

---

4. Configure Environment Variables

Create a .env file in the project root:

GROQ_API_KEY=your_groq_api_key

Never upload the .env file to GitHub.

---

5. Run the Application

streamlit run app.py

The application will open in your browser.

Note: Keep your GROQ_API_KEY private and never commit .env or database files.

---

🔑 Environment Variables

Variable	Description
GROQ_API_KEY	API key used to access the Groq API

---

🧠 Why Smart Mode?

Traditional chatbot applications often require users to manually choose a category before asking a question.

MultiGenAI removes that unnecessary step.

Traditional Approach

User
 ↓
Select Domain
 ↓
Ask Question
 ↓
AI Response

MultiGenAI Smart Mode

User
 ↓
Ask Anything
 ↓
AI Understands Intent
 ↓
Builds Context
 ↓
Generates Response

This creates a more natural and user-friendly AI experience.

---

📐 Design Philosophy

MultiGenAI follows a modular architecture.

Different responsibilities are separated into different modules:

Authentication
       ↓
Database
       ↓
Intent Detection
       ↓
Prompt Building
       ↓
LLM Generation
       ↓
User Interface

This makes the application easier to:

• Maintain
• Debug
• Test
• Extend
• Improve
• Deploy

---

🚧 Current Limitations

• SQLite is currently used for persistent storage.
• Intent detection depends on an LLM.
• Groq API usage is subject to model and service limits.
• Conversation memory is limited by the available context.
• No external document RAG system is currently integrated.
• No autonomous AI agent system is currently integrated.

---

🔮 Future Improvements

Possible future improvements include:

• PostgreSQL or MongoDB integration
• Google OAuth authentication
• Advanced long-term memory
• File upload and document understanding
• RAG-based knowledge retrieval
• Multi-model selection
• Voice interaction
• Image understanding
• Tool calling
• AI agents
• Personalized user preferences
• Advanced analytics dashboard

---

🌟 Future Vision

MultiGenAI is designed as a foundation for a more advanced AI ecosystem.

                    MultiGenAI
                        │
        ┌───────────────┼───────────────┐
        │               │               │
      GenAI            RAG            Agents
        │               │               │
   Generation      Knowledge       Autonomous
                    Retrieval        Tasks
        │               │               │
        └───────────────┼───────────────┘
                        │
                  Intelligent AI
                     Workspace

---

👨‍💻 Author

Subham Das

B.Tech Computer Science & Engineering

Areas of Interest

• Artificial Intelligence
• Machine Learning
• Deep Learning
• Natural Language Processing
• Generative AI
• Retrieval-Augmented Generation
• AI Agents
• Data Science

---

🎯 Project Goal

The goal of MultiGenAI is to build a practical, intelligent and extensible Generative AI workspace that understands what users actually want instead of forcing them to manually configure how the AI should respond.

---

📜 License

This project is created for educational, portfolio and development purposes.