#llm

Building a Kubernetes docs assistant that refuses to guess

You can wire up a document chatbot in an afternoon: split some files, embed them, stuff the top matches into a prompt, and let the model talk. It demos nicely. It also invents answers with confidence