Bigeye, the AI Trust Platform for the enterprise, is launching Policy Engine in Agent Trust Hub, enabling enterprises to control what their AI agents are allowed to do at runtime.
It also introduces an integration framework that lets AI gateway providers bring Bigeye's data context into their own policy decisions. Kong, TrueFoundry, OpenLLM, Bifrost (by Maxim AI) Solo.io, and Gravitee are the inaugural partners, according to the company.
Agent Trust Hub gives teams a registry of their agents and maps each one to the data it touches, with quality, classification, lineage, and governance context. Policy Engine turns that context into a decision each time an agent acts. Identity checks confirm that an agent may call a tool. Policy Engine adds whether the dataset behind that tool failed a freshness check an hour ago or holds fields never approved for AI use, the company said.
For every agent action, Policy Engine weighs identity, purpose, the requested action, and the current state of the data, then allows, denies, restricts, or escalates it.
"Agent Trust Hub showed teams what their agents do and whether the data behind them is trustworthy," said Eleanor Treharne-Jones, CEO of Bigeye. "Policy Engine makes it a guardian for every agent action, enforced at runtime through the gateways our customers already run."
AI gateways resolve each agent request to a specific tool, dataset, or field. Policy Engine brings the current state of that data into the gateway's decision, the gateway enforces the result, and every decision is recorded in one audit trail. The integration framework defines how a gateway connects to Policy Engine, so additional providers can integrate the same way the inaugural partners have. That gives enterprises an independent guardian layer for their agents, one that works across gateways and environments, said Bigeye.
Policy Engine and gateway integrations are available now for Agent Trust Hub customers.
For more information about this news, visit www.bigeye.com.