How many success stories with RAG

I thought to hear from practitioners, how many of you have been able to achieve accuracy of more than 70% with RAG application for Q&A on more than 10,000 documents. Have you seen context overlap as your document volume increases?