Google unveiled Gemini Enterprise Agent Platform at Cloud Next, targeting IT teams for agent-building with multi-model support including Anthropic's Claude.
At Google Cloud Next, CEO Sundar Pichai announced Gemini Enterprise Agent Platform — Google's direct answer to Amazon Bedrock AgentCore and Microsoft Foundry. The platform is aimed at IT and technical teams for building and managing agents at scale. Business users are directed to the separate Gemini Enterprise app for lower-code tasks like scheduling and file editing. Notably, the platform supports not just Google's Gemini and Nano Banana 2, but also Anthropic's Claude Opus, Sonnet, and Haiku — including the just-launched Claude Opus 4.7.
The platform's multi-model support — Gemini and Claude Opus/Sonnet/Haiku in one managed environment — means developers building enterprise agents no longer need to stitch together separate API providers or manage their own routing logic. The IT-first framing signals that Google expects developers, not business users, to own agent architecture decisions. This is a managed orchestration layer that could reduce significant infra overhead, but lock-in risk is real.
If your team is evaluating agent orchestration platforms this sprint, benchmark Gemini Enterprise Agent Platform's Claude integration against your current LangChain or custom routing setup — specifically check whether managed multi-model switching reduces your p95 latency on tool-calling tasks.
Suggested exercise · not independently tested. Access requirements and costs may vary.
Navigate to cloud.google.com and open the Vertex AI Agent Builder section
Create a test agent using the Gemini 1.5 Pro model, then switch the model selector to Claude Sonnet and run the same tool-calling prompt: 'Search for the top 3 recent news stories about enterprise AI and summarize each in one sentence'
Compare the two outputs side-by-side for quality, latency indicator, and token cost shown in the console
Tags