Tool use

Definition

The ability of an AI model to invoke external tools, such as web search, code execution, or database queries, to augment its capabilities. Tool use transforms models from knowledge stores into action-taking agents.

Why it matters

Tool use is the bridge between knowing and doing. A model without tools can only generate text based on its training data; a model with tools can search the web for current information, run code to verify its math, query databases for real-time data, and trigger real-world actions through APIs. This capability expansion is what makes agents possible and what turns AI from a content generator into a business process automator. For product developers, tool design is now as important as prompt engineering: the tools you give a model, and how you describe them, determine what the AI can accomplish. Poorly designed tool interfaces lead to misuse; well-designed ones unlock capabilities the model was already capable of but had no way to execute.

In practice

Anthropic's Claude supports tool use through a structured JSON interface where developers define tools with descriptions and parameter schemas. OpenAI provides a similar function calling API. In practice, common tools include: web search (for current information), code execution (for calculation and verification), database queries (for structured data retrieval), file operations (for document processing), and API calls (for external service integration). MCP standardizes tool interfaces across providers. The emerging best practice is to keep tool descriptions concise and specific, provide clear examples of when to use each tool, and implement error handling for failed tool calls.