7 Steps to Successful RAG at Enterprise Search & Discovery

Jeff Fried, InterSystems and self-described data management nerd, explained what he sees as the 7 steps to successful RAG. RAG can help with errors in search results because of outdated information and hallucinations. Common optimization areas for RAG are the data enrichment and staging, retrieval pipeline optimization, and reranking.

• Understand the problem
• Collect/prepare AI Ready data
• Chunk and embed effectively
• Adopt multiple retrieval techniques
• Implement prompts & guardrails
• Evaluate and improve
• Deploy and monitor

Lubor Ptacek, Egnyte, covered RAG for Real Business Problems and showed several demos of business-centric AI. He advised not to take content to AI. Instead, take AI to content. You should also think about data security, content intelligence, the user experience, and extensibility. He showed how to build your own agent, based on a Spanish translator template, that translates text into Czech. The more specific the problem, the better the success rate.

KMWorld returned to the J.W. Marriott in Washington D.C. on November 17-20, with pre-conference workshops held on November 17.

KMWorld 2025 is a part of a unique program of five co-located conferences, which also includes Enterprise Search & Discovery, Enterprise AI World, Taxonomy Boot Camp, and Text Analytics Forum.