Broadcom Releases AgentMinder to Check AI Agent Tool Calls at Runtime

Broadcom has made AgentMinder generally available, offering enterprises a control layer that checks an AI agent’s requested actions before they reach company resources. The product was introduced Aug. 31 at VMware Explore 2026.

The system is aimed at agents that do more than produce text. Broadcom says AgentMinder verifies an agent’s identity and authorizes every action against its declared mission, intent, context and current risk before access to an enterprise resource. That puts the check at the point where an agent requests data, invokes a tool or seeks to change a system.

Broadcom’s design treats an agent as more than a software identity. It binds an agent’s authority to a declared mission, permitted intents, approved tools and authorized resources. The agent must declare what it is trying to do before it can touch enterprise systems, according to the company.

• Runtime enforcement: a cloud-native AI gateway authenticates tokens, directs traffic only to authorized backends and uses a dynamic policy engine to assess context for each tool call.
• Observability and audit: an OpenTelemetry-based layer records agent sessions and actions, with chain-of-custody and anomaly-detection capabilities.
• Deployment: AgentMinder can sit beside large language models running on premises, in virtual private clouds or across public clouds. Broadcom says AuthZEN integration lets organizations reuse existing authorization endpoints.

Broadcom is also using its own deployment to make the scale case. The company says its AgentMinder-powered platform supports more than 20 million customer identities and 72,000 workforce identities. It says the platform handles nearly 36 million customer-related API calls and seven million workforce-related API calls each day through a multi-region, active-active architecture.

Those are company-supplied operating figures for Broadcom’s own platform. The product’s practical promise is broader: Broadcom says its combination of intent governance, runtime enforcement and auditability can let enterprises use agents in finance, HR and IT without rewriting their existing identity or authorization stack.