Tool Use
An LLM's ability to call defined functions to fetch data or take actions.
Tool use (sometimes called function calling) is the mechanism by which an LLM extends beyond its training data — calling APIs, querying databases, running code, browsing the web. It's the foundation that turns a chatbot into an agent.
The model emits a structured request: "call the lookup_customer function with id=42." The runtime executes that function and returns the result. The model continues the conversation with the result available as context.
In 2026, most frontier models support tool use natively. The interesting differentiation is in tool schema flexibility, parallel tool calls, and how well the model recovers from tool errors. MCP (Model Context Protocol) has become the dominant standard for defining tool surfaces across model providers.
Get the weekly digest
New tools, reviews, and prompts every Friday.