Document Chat Assistant
Chat application that answers questions on a set of business documents in PDF, Excel and Word, with conversations grouped into projects, kept as history and shareable by link.
- Area
- AI
- Main stack
- Angular · OpenAI API · RAG · LLM
Key features
- • Upload and segmentation of PDF, Excel and Word files
- • Chat answers based on uploaded document content
- • Conversations grouped into projects with persisted history
- • Conversation sharing by link
- • One-time access code sent by email
About the project
Chat through which someone asks questions of a set of company documents in PDF, Excel and Word, and gets answers drawn from their contents. Uploaded files are segmented and indexed in PostgreSQL, and the chat queries a language model over the retrieved passages. Conversations sit in projects, are kept as history and can be shared by link; access is by one-time code sent by email. An Angular front end with server-side rendering sits on an Express back end whose eight-table schema is created at start-up rather than by migrations.
Technologies used
Angular 
OpenAI API 
RAG 
LLM 
Angular SSR 
Node.js 
Express 
PostgreSQL 
pkg