GLOSSARY
Tool Calling
How models use software: instead of only producing text, they emit structured requests to call calculators, search engines, APIs — then weave the results into an answer.
You give the model a list of tools with names, descriptions and parameter schemas; when a task needs one, the model returns a structured call (tool: weather, city: Tokyo) instead of guessing. Your code executes it and hands the result back, and the model continues — looping until the task is done. This loop is what upgrades a chatbot into an agent.
It is also the failure surface: models hallucinate parameters, call tools in the wrong order, and prompt-injection can ride in through tool results. Strict schemas, timeouts and human confirmation for destructive actions are the practical mitigations.