Defined term
Tool use
An LLM's ability to call external functions, APIs, or services within a generation step.
Tool use (also: function calling) lets a model invoke deterministic capabilities — read a database, write a record, call an API, run code — as part of its response. Tool use turns LLMs from chat into agents. Production tool use requires schema validation, allow-listing, rate limits, and audit logging on every call.
When it matters
When the model needs to read or write to external systems (CRM, database, email, calendar). Required for any agent that takes action rather than just answers.
Real example
A sales agent with 4 tools: search_account (HubSpot read), get_recent_activity (CRM events), draft_outreach (template engine), schedule_send (email API). Each tool has a schema, allow-list, rate limit, and audit log.
KPIs to watch
Tool-call success rate (>95% target), tool-selection accuracy (right tool for the step, >90%), action rollback rate (<2%).
Related terms
Agentic AI
AI systems that can plan, take multi-step actions, and use tools to complete tasks autonomously.
Function calling
Specific implementation of tool use where the model emits structured JSON calls to registered functions.
MCP (Model Context Protocol)
Open protocol for exposing tools, resources, and prompts to AI models in a standard way.
Autonomous agent
An AI agent that completes a defined task without per-step human input.
See it in action
We use this every week
Book a 30-min call and we'll walk you through how Tool use shows up in a real engagement we're running.
Book a 30-min call