Tool use (function calling)
In one line: When an AI model can call external functions - Search, calculator, database - Instead of just generating text.
What is Tool use (function calling)?
Tool use (also called function calling) is the mechanism by which an LLM can invoke external functions during a conversation, rather than just generating text. The model decides which function to call and with what arguments; the application runs it; the result flows back to the model as additional context; the model uses that context to produce its final answer. Tool use is what transforms a text-in / text-out model into a capable AI agent.
The tool use flow
- Define tools - The developer provides the model with a list of available functions: e.g.
search_web(query),get_stock_price(ticker),send_email(to, subject, body). - User sends a request - 'What is Apple's stock price today?'
- Model decides to call a tool - It returns a structured call:
get_stock_price(ticker='AAPL'). - Application executes the function - Your code runs the call and returns the result: 'AAPL: $212.45'.
- Model synthesises the answer - It uses the live result to respond: 'Apple (AAPL) is currently trading at $212.45.'
This pattern lets LLMs access live data, run precise calculations (avoiding LLM arithmetic errors), write to databases, send messages, and control software - Overcoming the core limitation of knowledge cutoff and static training data.
Common tools LLMs use
| Tool | What it does | Example |
|---|---|---|
| Web search | Retrieves live information from the internet | 'What happened in the news today?' |
| Code interpreter | Runs Python (or other) code and returns output | 'Calculate the compound interest on £10,000 at 4% for 5 years' |
| File reader/writer | Reads or writes files on the local system | 'Summarise all the PDFs in this folder' |
| Database query | Runs SQL or API calls against a data source | 'How many orders were placed last week?' |
| Browser/computer use | Controls a real browser or desktop GUI | 'Book me a meeting room on our internal portal' |
| Email/calendar | Sends messages or creates calendar events | 'Schedule a follow-up call with the client for next Tuesday' |
Tool use vs agentic loops
A single tool call is the basic case. When the LLM calls tools in sequence - Searching for information, using it to decide the next search, then synthesising the results into a report - That is a multi-step agentic loop. Tools are the primitive; agents are the loop that orchestrates them. Reasoning models excel in agentic settings because their extended chain-of-thought helps them plan which tools to use and in what order.
Standardisation with MCP
Model Context Protocol (MCP) is an open standard from Anthropic that creates a uniform way to define, host, and connect tools to any compatible LLM. Instead of writing custom integration code for every model, you build an MCP server once and it works with Claude, ChatGPT, and any other MCP-compatible client. Both the OpenAI and Anthropic APIs support tool use natively, but MCP provides the portable layer that is driving adoption of standardised agent tooling across the industry.
Tool use (function calling) example
If you are using AskAI.free, a practical way to understand tool use (function calling) is to ask a model to explain it, then ask for a concrete example in your own workflow. For example: "Explain tool use (function calling) for someone using AI to write, code, research, or create images."
This turns the term from a dictionary definition into a decision-making tool: you can see when it affects prompt quality, model choice, output reliability, privacy, cost, or how much context the AI can use.
Why Tool use (function calling) matters
Tool use (function calling) matters because it changes how you choose, prompt, compare or trust AI systems. If you understand this term, you can ask better questions, spot weak answers faster and choose the right model or tool for the job.
A common mistake is treating tool use (function calling) as isolated jargon. It usually connects to nearby ideas like AI agent and Alignment, so check those next if you want the full picture.
Common mistake with Tool use (function calling)
The most common mistake is using the term as a label without changing behavior. When tool use (function calling) comes up, ask what action should change: the prompt, the model, the input length, the evidence you request, or the way you verify the answer.
See it in action - Ask any AI about tool use (function calling) on AskAI.free.
Try it free →