AI Agent
A software system that uses an AI model to perceive its environment, make decisions and take actions to achieve a goal over multiple steps.
In detail
An AI agent combines a foundation model with the ability to use tools: calling APIs, reading files, browsing the web, querying databases, sending emails or triggering other software. The agent runs a loop: given a goal, it generates a plan, uses a tool, observes the result and decides what to do next. This continues until the goal is achieved or the agent stops. Frameworks like LangChain, LlamaIndex, the Vercel AI SDK and Anthropic's Claude with tool use all support building agents. The Model Context Protocol provides a standard way for agents to discover and call tools.
Why it matters for Australian business
AI agents are moving from demos to production across Australian businesses. Common deployments include email triage agents, CRM auto-update agents, scheduling agents and document processing agents. The risk profile differs from a simple chatbot: an agent that can take actions (send email, update records, call APIs) can make irreversible mistakes at machine speed. Well-designed agents have human-in-the-loop checkpoints for consequential actions, audit logs, and hard limits on what they can touch.