The Future of Enterprise AI Agents: Governance, Security, & Scale

Enterprise AI Agent Platforms

Michael Clark of Google Cloud discusses building and governing enterprise AI agents on the Gemini enterprise agent platform. He traces his path from privacy and discriminative AI at Meta to agent infrastructure, viewing agents as integrating the full technology stack.

Interoperability and Openness

He argues interoperability, openness, and transparency are essential for agent ecosystems. Connections via MCP, agent-to-agent protocols, legacy APIs, and skills determine whether recursive, self-improving agents can integrate into business processes.

Platform Architecture

Google's platform spans the Agent Development Kit, hosting runtimes, context tools, and dedicated governance plus optimization and evaluation layers. The same infrastructure supports first-party Google products and external enterprise developers.

Consumer vs. Enterprise Governance

Consumer and enterprise products face different data obligations—regulations like GDPR versus internally defined risk tolerance—but both require similar governance concepts. Clark notes enterprise release cycles differ from consumer products because of compliance and accountability requirements.