Projects/Chatbot for Customer Support (LLM)
AI / MLFinal Year

Chatbot for Customer Support (LLM)

Real-time AI chatbot with conversation memory and analytics.

Case Study

Problem Statement

Small businesses can't afford round-the-clock support staff, so common customer questions go unanswered outside working hours, hurting response time and satisfaction.

Our Solution

An LLM-powered chatbot answers FAQs using a knowledge base (RAG — retrieval-augmented generation), remembers conversation context, and logs every conversation for an analytics dashboard.

Features

  • RAG pipeline over a custom knowledge base (PDF/FAQ ingestion)
  • Conversation memory within a session
  • Escalation to a human/lead-capture form when the bot can't answer
  • Analytics dashboard: most-asked questions, unresolved queries
  • Embeddable chat widget for any website

Technology Stack

PythonLangChainOpenAI/Anthropic APIFAISS/Chroma vector DBFastAPIReact

How It's Built

  1. 1

    Documents are chunked and embedded into a vector store

  2. 2

    On each user message, relevant chunks are retrieved and passed to the LLM with the conversation history

  3. 3

    FastAPI backend serves the chat endpoint and logs each exchange

  4. 4

    React widget renders the chat UI and can be embedded via a script tag

FAQ

Do I need to pay for API usage?

The project supports both paid APIs (OpenAI/Anthropic) and free local models (via Ollama) — we'll help you pick based on your budget.