AI-powered mobile application to extract, structure, and understand information from unstructured documents across domains.
The mission of this project is to create an MVP that runs on mobile devices. Core functionality includes extracting structured information from uploaded documents, storing and organising it within a personal knowledge base, and enabling conversational querying using RAG with optional enrichment from external knowledge sources such as research papers. The system will demonstrate domain configurability using medical documents as the primary domain and finance as a secondary domain.
Important information is often locked in unstructured documents:
- Medical reports without clear interpretation
- Financial data spread across invoices and timesheets
- Legal documents requiring manual cross-referencing
AIlixir Intelligence solves this by making data structured, connected, and conversational.
- Frontend: React Native (Expo)
- Backend: Python + Google Cloud Run
- AI/ML: Vertex AI, Document AI
- Database: Firestore
- Knowledge Layer: Vector Search / Neo4j
View https://miro.com/app/board/uXjVHcF54Y8=/?share_link_id=142814163268
- Document scanning and upload
- AI-powered data extraction
- Domain-configurable architecture (e.g. medical, finance)
- Chat interface grounded in user data
- Integration of external knowledge sources
- Medical (lab reports, blood results)
- Finance (invoices, timesheets, expenses)
# clone repository
git clone <repo-url>
# navigate into project
cd <project-folder>
# install dependencies
npm install
# start Expo app
npx expo start