AI Agent

An AI agent receives a goal and breaks it into steps: it can read documents, call APIs, write code, or book meetings, depending on which tools it can access. Unlike a classic chatbot that only replies in one session, an agent usually tries to complete something over multiple steps.

Many modern agents are built on a Large Language Model (LLM) plus memory, planning, and tool access. Compare automation and prompt engineering, or see OpenClaw and n8n for different ways to build agentic workflows.

---

Key characteristics

• Breaks goals into sub-steps and can use tools, APIs, and memory to complete tasks.
• Differs from a regular chatbot by acting toward a goal over time, not just answering one prompt.
• Is most valuable when the task requires multiple decisions, external feedback loops, and error handling.