Tool Use (LLM)
LLM tool use is the pattern in which a language model acts through external functions, APIs, databases or code execution: the model requests a tool, the host application runs it and returns the result, and the model continues, turning a text generator into a controller that can act. Function calling is the structured-output mechanism most tool use is built on.
AIDOLS Research Team · Editorial standards
Full definition
The model is given a JSON schema describing available tools; at inference time it can produce a structured tool-call rather than (or in addition to) free text. The host application executes the tool and feeds results back into the model's context. Tool use is the foundation of RAG, agents, and most production LLM applications. Major implementations include OpenAI Function Calling, Anthropic Tool Use, and the open-source equivalents in vLLM and llama.cpp.
Why it matters
Tool use is what makes LLMs useful for real workflows. Without tools, a model can only describe what to do; with tools, it can do it — query the database, send the email, file the ticket, run the code. Tool design (which tools to expose, how to structure them) is now a primary skill for AI engineers.
Example
A travel-planning assistant exposes 5 tools: search_flights, search_hotels, get_weather, get_user_preferences, and book. The LLM orchestrates them in response to "plan me a 4-day trip to Tokyo for under $3,000."
Next steps
Related terms
- AI Agent / Agentic AIAn AI agent is an LLM-driven system that, given a goal, plans a sequence of steps, calls tools or APIs, observes the results, and iterates until the goal is reached — going beyond single-turn chat to multi-step autonomous action.
- Large Language Model (LLM)A Large Language Model (LLM) is a deep neural network — almost always a transformer — trained on hundreds of billions to trillions of words to predict the next token, and to generate, summarize, translate, or reason over text.
- Retrieval-Augmented Generation (RAG)Retrieval-Augmented Generation (RAG) is a technique where a language model retrieves relevant documents from an external knowledge base before generating a response, which reduces hallucination and enables citation of sources.
- Context WindowA context window is the maximum amount of text — measured in tokens — a language model can read and reason over in one inference call, equivalent to the model's working memory for that turn.
More deployment & operations terms
- Knowledge GraphA knowledge graph is a structured representation of entities (people, products, customers, drugs) and the relationships between them, stored as a graph and used to ground AI systems in verified, queryable facts rather than free-form text.
- Feature StoreA feature store is a centralized data system that stores, versions, serves, and reuses curated machine-learning features — guaranteeing the same feature definitions are used at training time and at online inference time.
Source & further reading
Primary source: OpenAI — "Function calling and other API updates" (2023).
Citation policy: this entry is part of the AIDOLS AI Implementation Glossary and may be quoted for research, journalism, and education with attribution to aidolsgroup.com/en/glossary/tool-use/.