Google pitches governed agent stack
- Google Cloud used its April 22 Next keynote to launch Gemini Enterprise Agent Platform, recasting Vertex AI as a governed system for building and running agents. - Google tied the platform to Oracle data and ServiceNow workflows, adding natural-language database access, orchestration, security controls, and cross-vendor agent handoffs. - The pitch extends Google’s broader “agentic enterprise” push unveiled at Next in Las Vegas. (cloud.google.com)
Google Cloud used its April 22 Next keynote in Las Vegas to pitch a simple idea: companies should run artificial intelligence agents on a centrally governed platform, not as scattered experiments. (techrepublic.com) (cloud.google.com) The centerpiece was Gemini Enterprise Agent Platform, which Google described as the evolution of Vertex AI into a system to build, scale, govern, and optimize agents. Google also folded the Gemini Enterprise app and the new platform into a broader Gemini Enterprise package for employees and developers. (cloud.google.com 1) (cloud.google.com 2) In plain terms, Google is selling agents as software workers that can take actions across apps and data, while information technology teams keep the permissions, security, and audit controls in one place. Thomas Kurian framed that as the operating model enterprises need as agentic AI moves from pilots into production. (techrepublic.com) (cloud.google.com) Google paired that message with partner integrations. Oracle said on April 22 that its Oracle AI Database Agent for Gemini Enterprise lets customers query Oracle AI Database@Google Cloud in natural language, without writing SQL. (oracle.com) (blogs.oracle.com) Google and Oracle also said Google’s Knowledge Catalog is being extended to Oracle AI Database@Google Cloud in preview, creating a shared metadata and governance layer so agents can find and use Oracle data with more context. (cloud.google.com) (oracle.com) ServiceNow announced the same day that its agents and Google Gemini Enterprise agents can work in one chain across 5G operations, retail, and information technology systems. The companies said the expanded partnership is aimed at autonomous enterprise operations rather than single-app assistants. (newsroom.servicenow.com) That combination points to Google’s actual sales pitch: keep the data governed, let employees ask questions in everyday language, and let multiple vendors’ agents pass work between systems. Google used Next to position itself as the control plane for that model, even when the underlying data or workflows live with Oracle or ServiceNow. (cloud.google.com 1) (cloud.google.com 2) (newsroom.servicenow.com) Google wrapped the software message in scale. In its Next recap, the company said the conference drew more than 32,000 attendees and included more than 250 product, customer, and ecosystem announcements. (cloud.google.com) The company also put money behind the ecosystem. Multiple reports from Next said Google launched a $750 million partner fund alongside the Gemini Enterprise Agent Platform and new infrastructure announcements. (crn.com) (msn.com) The thread running through the week was less about one chatbot and more about who manages the stack. Google’s answer was that agents, data access, orchestration, and guardrails should sit inside one enterprise platform, even if the work spans several vendors. (infoworld.com) (cloud.google.com)