Tool Use
Letting a model reach outside its own head — run code, search the web, hit an API — instead of guessing from memory.
Think of it like
A smart person who, instead of estimating, just picks up a calculator, opens a browser, or makes a phone call.
Example
Asked for today’s weather, the model does not invent a forecast — it calls a weather API and reads back the real number.
How it actually works
The model does not run tools itself; it emits a request to use one, a surrounding program runs it, and the result comes back as new text in the context. This turns a frozen, all-knowing guesser into something that can act and observe. The whole agent field is basically tool use in a loop.
For product teams
The single feature that turns a chatbot into something that can actually get things done.
For engineers
The model outputs a tool invocation; an external runtime executes it and feeds the result back into the context window.
Related
- Function Calling — The structured version, where the model fills in named arguments.
- Agent Loop — A tool call plus its returned observation, run over and over.
- Hallucination — It reduces made-up answers by fetching real ones.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome