Develop a RAG system for our internal knowledge baseExample listing

We are a 70-person professional services company with a significant amount of internal documentation including company policies, client SOPs, product and service specifications, and training materials, totaling roughly 400 documents in a mix of PDF and Google Docs format.

We want our team to be able to ask natural language questions and get accurate, cited answers drawn from these documents instead of manually searching through folders. We are not looking for a general-purpose chatbot, we specifically want a RAG system that only answers from our document corpus and clearly cites its sources.

Preferred infrastructure would be something we can run cost-effectively, either a hosted solution with low ongoing cost or a self-hosted option on our existing cloud account (AWS). We use Notion and Google Workspace heavily, so integration with those as document sources would be a significant plus.

The scope covers: document ingestion pipeline, embedding and vector store setup, retrieval and generation layer, and a simple web UI for staff to use. We are open to architecture proposals, this is partly why we have set the budget as open to offers.