Glossary

What is Tool Use in AI?

Tool use — also known as function calling — is the ability of AI models to invoke external tools, functions, and APIs to accomplish tasks beyond their native capabilities. It's what separates AI chatbots (which can only talk) from AI agents (which can act).

How Tool Use Works

User: "What's the weather in Berlin?"

AI Model:
  1. Analyzes request → needs current weather data
  2. Selects tool: get_weather(location: string)
  3. Outputs structured call:
     { "tool": "get_weather",
       "args": { "location": "Berlin" } }

Application:
  4. Executes get_weather("Berlin")
  5. Returns: { "temp": 22, "condition": "sunny" }

AI Model:
  6. Incorporates result
  7. Responds: "It's 22°C and sunny in Berlin."

This loop — reason → select tool → execute → incorporate result — is the fundamental cycle of AI tool use. Complex tasks may involve dozens of tool calls in sequence.

Function Calling vs MCP

AspectFunction CallingMCP Tool Use
ScopePer-model (GPT, Claude separately)Universal (any MCP-compatible model)
DiscoveryManual (developer hardcodes tools)Automatic (server advertises tools)
SecurityDeveloper's responsibilityProtocol-level (scanning, sandboxing)
DistributionCustom (per application)Marketplace (SkillExchange)
InteroperabilityLow (vendor lock-in)High (open standard)

Common Tool Categories

🔍 Search & Research

Web search, document search, knowledge base queries

💻 Code Execution

Run Python, JavaScript, SQL, shell commands

📊 Data Processing

Parse files, transform data, generate charts

🌐 Web Interaction

Browse websites, fill forms, scrape pages

📧 Communication

Send emails, post to Slack, create tickets

🗄️ Storage

Read/write files, databases, cloud storage

Why Tool Use Matters

Without tool use, AI models are limited to their training data — they can talk but can't act. With tools, AI becomes agentic: it can fetch real-time data, execute code, interact with external systems, and accomplish real work. Tool use is the bridge between AI intelligence and real-world impact.

Read more: What is an AI Agent? · What is MCP? · What is Agentic Commerce?

FAQ

What is tool use in AI?

Tool use (also called function calling) is the ability of an AI model to invoke external functions, APIs, or tools to accomplish tasks it couldn't do alone. For example, an AI using a calculator tool to do math, or a web search tool to find current information.

How does function calling work?

The AI model is given a list of available functions with their schemas (name, parameters, description). When the model decides a function is needed, it outputs a structured call with the function name and arguments. The application executes the function and returns the result to the model.

What's the difference between function calling and MCP?

Function calling is a capability of individual LLMs — each model (GPT, Claude, Gemini) has its own implementation. MCP (Model Context Protocol) standardizes this at the protocol level: any MCP-compatible model can use any MCP tool. MCP adds discovery, security, and universal compatibility.

What types of tools can AI agents use?

Agents can use virtually any digital tool: calculators, web browsers, database queries, code execution, file operations, API calls, image processing, email sending, deployment tools, and more. On SkillExchange, you'll find hundreds of MCP tools across all categories.

How do I make my tool available to AI agents?

Package your tool as an MCP server and publish it on SkillExchange. Our platform handles discovery, security scanning, payments, and distribution. Any MCP-compatible AI agent (Claude, GPT, Gemini, custom agents) can then use your tool.

Build tools for AI agents

Publish your MCP tools on SkillExchange and reach thousands of agents.