An AI assistant waits for a person to ask and then helps: it answers a question, rewrites a paragraph, summarizes a document or explains some code. The person stays in charge of each step.
That is the main difference from an AI agent, which is given a goal and decides its own steps. The line is blurry because many assistants can now use tools, so a good test is who drives the work. If you prompt each step, it is an assistant. If you hand over a goal and it runs a loop by itself, it is behaving as an agent.
Assistants are built on a large language model and usually shaped by a system prompt that sets their tone, scope and limits.
Related: Generative AI, Agentic AI, Prompt Engineering.