Enterprise-ready document processing for AI

In this talk, Tu and Kerim look at what they learned about building an enterprise-ready document processing pipeline to make an internal chatbot smarter and capable of providing actually relevant answers, all the while safely respecting privacy, by sanitizing information that shouldn't be immediately visible. The speakers look at how documents should be prepared, how they can be ingested at rapid scale, and where the PII sanitization needs to be applied. Attendees can expect to learn actionable strategies and will get access to a repository they can clone and get started with, irrespective of the underlying model they prefer.