🚀 Doc Analyzer - AI Multi-Document Chat Assistant

Inspiration

I was inspired by the challenge of managing and extracting insights from multiple documents simultaneously. Students, researchers, and professionals often struggle to analyze large volumes of PDFs and images efficiently. I wanted to create an intelligent assistant that could understand multiple documents at once and provide instant answers.

What it does

Doc Analyzer is an AI-powered document chat assistant that allows users to:

• Upload multiple documents (PDFs, images) in one go
• Automatically process and analyze all documents using advanced AI
• Generate comprehensive summaries of document collections
• Ask natural language questions and get accurate, context-aware answers
• Receive AI-generated suggested questions for deeper insights
• Maintain conversation history across sessions

How I built it

• Frontend: Gradio with custom CSS
• AI Processing: LangGraph Agentic workflow for multi-file document orchestration deployed on AWS Bedrock AgentCore Runtime leveraging it's memory primitive
• Document Handling: Vector embeddings for semantic search and retrieval
• State Management: Session-based state persistence for user continuity
• Architecture: Modular agent-based system with separate processing pipelines for different document types

Built With

• agenntcore
• amazon-web-services
• bedrock
• chromadb
• gradio
• langgraph
• llms
• python
• textract