What is AI agent?
An AI system that runs autonomously, uses tools (shell, APIs, files), and pursues multi-step goals without a human in the loop for each action. Distinct from a chatbot, which only replies.
Workflows versus agents
Anthropic's engineering guide on building effective agents separates two kinds of agentic system. In a workflow, LLMs and tools are orchestrated through predefined code paths. In an agent, the LLM directs its own process and tool use, deciding what to do next as it goes.
The building block
The guide calls the basic unit an augmented LLM: a model enhanced with retrieval, tools and memory. Current models can use these themselves, writing their own search queries, choosing tools and deciding what to keep.
When to use one
The guide advises starting with the simplest approach that works, because agentic systems usually cost more in latency and money. Workflows suit well-defined tasks; agents suit open-ended problems where flexibility and model-driven decisions matter. For many applications, a single well-optimised model call with retrieval and examples is enough.
Example
A scheduled routine that checks release feeds, decides which pages are affected, edits them and opens a commit is an agent: the model chooses the steps. A script that always summarises the same feed with a fixed prompt is a workflow.
Sources
See also
← Back to the full AI agent glossary.