LLM Function Calling

Definition

A capability where language models can generate structured function call requests — specifying which tool to invoke and what parameters to pass — rather than just generating text. Function calling enables AI agents to interact with external systems: databases, APIs, file systems, and other tools. Supported by OpenAI, Anthropic, Google, and most major model providers.

Why It Matters

Function calling is the bridge between AI reasoning and real-world action. When a model generates a function call, that call can modify production state — delete records, send emails, transfer funds. The model generates the call, but the system executes it. There is typically no validation layer between generation and execution.

How Exogram Addresses This

Exogram sits between function call generation and execution. Every call the model generates passes through the deterministic policy engine before reaching your tools. Schema validation, intent analysis, boundary control, and loop protection — all in 0.07ms. Works with OpenAI, Anthropic, Google, and any provider's function calling API.

Related Terms

medium severityProduction Risk Level

Key Takeaways

  • This concept is part of the broader AI governance landscape
  • Production AI requires multiple layers of protection
  • Deterministic enforcement provides zero-error-rate guarantees

Governance Checklist

0/4Vulnerable

Frequently Asked Questions