Twine

AskDoc: Flutter AI Document Chat App (RAG)

MUHAMMAD USMAN RANDHAWA
Published

0
  • Project creator

AskDoc is an AI document chat app I built with Flutter for Android and iOS, powered by Google Gemini AI.

The goal was to let anyone upload a PDF, receipt or photo and simply ask questions about it, instead of reading pages to find one answer.

What I built:
• Chat with any PDF: ask questions and get precise answers from the document
• Instant AI summaries: key points in seconds, like a three-bullet summary
• Scan and understand receipts, invoices, menus and photos with AI vision
• Add documents from files, camera or gallery
• Document library with search, history and copy or share answers

Tech stack: Flutter, Dart, Google Gemini API (text and vision), RAG (retrieval-augmented generation), secure backend for the API key.

Challenge: keeping answers accurate and grounded in the uploaded document. I built a RAG pipeline that splits documents into chunks, finds the most relevant parts for each question and sends only those to Gemini, so answers stay precise and costs stay low.

Result: a smart, polished AI assistant app that shows how I build chatbots, RAG and AI vision with Gemini, ChatGPT or any AI model, ready to adapt for legal, education, finance, HR or customer support.