Tool use / function calling
Tool use
The LLM emits a structured request to call an external function (search, calculator, API), the host runs it, the result goes back in the next turn. Foundation of every agent worth shipping.
Tool use formalises "LLM asks for help from the outside world." The model emits a JSON object (or proprietary format) specifying which tool to call and with what arguments; the host runtime parses it, executes the tool, and includes the result in the next prompt.
OpenAI calls it function calling; Anthropic calls it tool use; both are the same concept. Quality has climbed steeply: 2023-era function calling was unreliable enough to need fallback parsers; 2026 frontier models handle 50+ tools, parallel calls, and structured outputs cleanly.
Common tools: web search, code execution, database queries, calendar/email reads, file system access. MCP is the standardisation layer on top.
See also
-
MCP
Model Context ProtocolOpen standard from Anthropic (Nov 2024) for connecting AI assistants to external tools, data sources, and prompts. Think "USB for LLM tools" — one protocol, many servers, any client.
-
Agentic
Agentic systemsLLM-driven loops that plan, take actions in the world (call tools, edit files, hit APIs), observe results, and iterate — rather than just answering a single prompt. The dominant 2026 paradigm for AI engineering.