AI agent

Also: LLM agent, autonomous agent

Wraps a language model in a loop with tools, memory and a stopping rule: at each step it decides on an action (a tool call or a final answer), the runtime executes it, and the observation is fed back. ReAct, which interleaves thoughts and actions, is a common pattern.

Example

Asked to book dinner in Paris if the weather is good, a ReAct agent calls weather_api.get("Paris") → "Sunny, 22°C", then restaurant_api.search("Paris", "outdoor") → 5 results, then books Le Jardinet for 7 PM.

See it in the lesson: Agent Architectures →

Related terms

More in Retrieval, RAG & Agents