AI Document Intelligence
Search, summarize and ask questions about documents
Find answers in document collections and follow the references back to the original files.
- Project type
- Python document application with a browser interface.
- My role
- Application development: document ingestion, semantic retrieval, question answering and the Flask interface.
- What it delivers
- Index PDF, Word and text files, select documents, view summaries and ask questions with clickable source references.
- Python
- Flask
- SentenceTransformers
- ChromaDB
- SQLite
Explore the prepared walkthrough, or email me to arrange a live demonstration of document search, summaries and source references using sample documents. The demo request opens your email app; we agree a time by email.
Hosting, AI processing & access
Hosting: The demonstration application stores its documents, search index and indexing records on the machine where it runs. Hosting for a client system is agreed around the project's requirements.
AI processing: When OpenAI answers are enabled, the application sends your question and retrieved document excerpts to OpenAI. Without an OpenAI API key, it returns extracted passages with source references.
Access: Selecting documents controls the search scope. A shared deployment also needs authentication and user permissions appropriate to the organisation; these must be implemented as part of that deployment.
Before using confidential documents, we agree which sources and external services may be used, who can access the system, and the requirements for storage, backups and retention.