AI-Support-Assiistent-

A lightweight RAG-based (Retrieval Augmented Generation) Support Assistant built using Groq LLM + ChromaDB + LangChain + Streamlit.

The agent answers user queries using your FAQ data, ensuring accurate and context-based responses.

---

📌 1. Overview

This AI Support Assistant helps users get automated answers to common support questions.
It uses:

• A vector database (ChromaDB) to store & retrieve FAQ data
• SentenceTransformer embeddings to match similar questions
• Groq LLM (llama-3.3-70b-versatile) for fast, accurate responses
• Streamlit for the UI

This makes the tool behave like a mini customer support agent.

---

🚀 2. Features

✔ Context-aware answers

Uses RAG to answer strictly from provided FAQ data.

✔ Fast inference with Groq

Groq’s hosted LLM provides extremely high speed and low latency.

✔ Expandable FAQ database

Simply update the faqs.txt file to add new knowledge.

✔ Clean UI using Streamlit

Easy for users to interact and test the assistant.

---

⚠️ 3. Limitations

• Cannot answer questions outside FAQ context.
• No authentication (anyone can access if deployed publicly).
• Basic UI (no chat history).
• Not suitable for very large knowledgebases without optimization.

---

🧠 4. Architecture Diagram

                 ┌──────────────────────────┐
                 │        User (UI)         │
                 │  Streamlit Frontend      │
                 └─────────────▲────────────┘
                               │  User Query
                               │
                 ┌─────────────┴────────────┐
                 │      SupportAgent         │
                 │ (Controls full pipeline)  │
                 └─────────────▲────────────┘
                               │
            ┌──────────────────┼──────────────────┐
            │                  │                  │
            │                  │                  │
            ▼                  │                  ▼
  ┌──────────────────┐         │        ┌──────────────────────┐
  │  ChromaDB Vector  │         │        │ SentenceTransformer │
  │     Database      │◄────────┘        │  Embedding Model     │
  └──────────────────┘   Similarity      └──────────────────────┘
        ▲   │              Search                ▲
        │   │                                    │
        │   └───────────────┬────────────────────┘
        │                   │
        │       Relevant FAQ Snippets
        │                   │
        ▼                   │
   ┌──────────────────────────────────────────────┐
   │       Groq LLM (llama-3.3-70b-versatile)     │
   │  Generates the final answer using context     │
   └──────────────────────────────────────────────┘
                               │
                               ▼
                  ┌──────────────────────────┐
                  │     Final Answer         │
                  │  Shown in Streamlit UI   │
                  └──────────────────────────┘

🛠️ 5. Tech Stack & Models Used 🔹 Backend & AI

Groq LLM: llama-3.3-70b-versatile

SentenceTransformer: all-MiniLM-L6-v2 for embeddings

LangChain for text splitting & retrieval pipeline

ChromaDB for vector storage

Python 3.10

🔹 Frontend

Streamlit (interactive user interface)

🔹 Other Tools

dotenv for API key management

pandas for handling text data

GitHub for version control

Streamlit Cloud for deployment

⚙️ 6. Setup & Installation Instructions

1. Clone repository

git clone https://github.com//AI-Support-Assiistent-.git cd AI-Support-Assiistent-

2. Create virtual environment

python -m venv venv venv\Scripts\activate # on Windows

OR

source venv/bin/activate # Mac/Linux

3. Install required packages

pip install -r requirements.txt

4. Create .env file

touch .env

Add inside .env:

GROQ_API_KEY=your_api_key_here

5. Run the app

streamlit run app.py

7. Project Structure (Required) AI-Support-Assiistent-/ │ ├── app.py # Streamlit UI ├── agent.py # Main support agent logic ├── retriever.py # RAG system (Chroma + Embeddings) ├── list_models.py # Utility to list Groq models │ ├── data/ │ └── faqs.txt # FAQ knowledge base │ ├── requirements.txt # Python dependencies ├── .env.example # Example env file ├── .gitignore └── README.md

8. Potential Improvements Future Enhancements

Add chat history for multi-turn conversations

Add authentication for restricted access

Add admin panel to upload FAQs dynamically

Use bigger embedding models for better retrieval

Add feedback system (thumbs up/down)

Improve UI design with animations + branding

Deploy using Docker or HuggingFace Spaces

9. Demo Link 🔗 Live Demo: https://