Answer

RAG: How AI Looks Stuff Up Before Answering

Most AI chatbots only know what they learned during training, like a student who studied months ago and then took a test with no notes. RAG — Retrieval-Augmented Generation — gives the AI a open-book exam instead. Before answering your question, it searches a pile of documents, grabs the most relevant pieces, and uses those to build its answer. This means the AI can work with fresh information, your company's private files, or anything it was never trained on. It reduces the AI making stuff up because it is anchoring answers to real sources you can check.

Example: A customer service bot powered by RAG searches your company's latest return policy PDF before answering 'Can I return this after 60 days?' — instead of guessing based on old training data.