There’s AWS Bedrock Knowledge Base (Amazon proprietary RAG solution) which can digest PDFs and, as far as I tested it on real world documents, it works pretty well and is cost effective.