Answer
Building AI agents with LLMs means using large language models as the core reasoning engine that decides what steps an agent should take to complete a task.
Real-world example A developer builds a research agent using an LLM combined with a search tool to automatically gather and summarize information on a topic.