Tool Calling
A capability of modern LLMs where the model emits structured JSON requesting that the host program execute a named tool with arguments — used to give the LLM internet, code execution, database, or any external action.
Tool calling (also "function calling") is the foundation of agentic AI. The host registers a JSON Schema for each tool; the LLM, given a user prompt, may respond with a tool_use block specifying the tool name and arguments. The host runs the tool, returns the result as a tool_result block, and the LLM continues. Modern Claude and GPT support parallel tool calls (multiple tools in one turn) which dramatically reduces latency.
Related terms
AI systems where an LLM plans and executes multi-step tasks by calling external tools, accessing files, browsing, and adjusting its own approach based on results.
Anthropic's official Python and TypeScript SDK for building production agents with Claude — handles the tool-use loop, hooks, and slash commands.
A neural network with billions of parameters trained on broad text corpora to predict and generate language — the engine behind ChatGPT, Claude, and Gemini.
Need this built into a real product?
Viprasol Tech ships production code for everything defined here — MT4/MT5 EAs, AI agents, B2B SaaS, AWS architecture.
Send a brief →