Function calling is how an AI asks software to run a specific action with specific inputs. The model returns a structured request that matches a defined schema, and the surrounding software decides whether to run it or reject it. That wrapper can require certain fields or block certain values, which is why structured data makes agents safer. It’s the machinery behind a tool call in Claude Code, Codex, or an agent log.
A work order form makes this concrete. “Can you handle the thing for the client?” is vague. A form with client name, action requested, due date, and approval status tells the next person exactly what to do. Function calling turns a loose request into that usable instruction: the AI picks the action it needs and fills in the form.
Why you care
For a nontechnical operator, the question is, “What forms should the AI be allowed to submit?” A model might call search_transcripts with a client name and date range. The names matter less than the pattern: defined action, defined inputs, structured output. Define the forms well and you give the agent controlled action instead of open-ended authority.