PROJECT
AI Doc Chat
An AI-powered documentation chat application using RAG (Retrieval-Augmented Generation) to provide context-aware answers from uploaded documents.
Project Overview
AI Doc Chat is an advanced documentation chat application built with RAG architecture. It allows users to upload documents (PDF, Markdown, Text) and ask questions about them. The system parses, chunks, and embeds the document content into ChromaDB. When a question is asked, it retrieves the most relevant chunks using semantic search and uses Google Gemini 2.5 Flash to generate accurate, context-grounded answers, significantly reducing AI hallucinations.
Key Features
Document Upload & Parsing
Support for various document formats including PDFs, Markdown, and text files with intelligent text extraction.
Intelligent Chunking
Smart document chunking with overlap to preserve context across boundaries and improve retrieval accuracy.
Semantic Retrieval
Advanced semantic search using Gemini Embedding model and ChromaDB for highly relevant context retrieval.
AI Chat Interface
Interactive chat UI with source attribution, showing exactly which document sections informed the answer.
Processing Status Tracking
Real-time visual feedback on document processing states (parsing, embedding, ready).
Persistent State
Maintains chat history and document state across sessions for continuous workflow.
Error Handling
Robust error handling and fallback responses to gracefully manage failures or unanswerable queries.