Google takes a different approach with its new enterprise AI agent-building tool

Google CEO Sundar Pichai opened the Google Cloud Next conference on Wednesday with a video unveiling one of the company's major new products: the Gemini Enterprise Agent Platform.

The tool is designed to build and manage AI agents at scale, positioning itself as Google's response to Amazon's Bedrock AgentCore and Microsoft's Foundry.

With AI agents advancing fastest in technical areas such as coding, and enterprise adoption still facing security concerns due to the technology's early stage, Google has taken a distinct approach with this product. The Gemini Agent Platform is primarily aimed at IT and technical teams within organisations.

At the same time, business users are directed toward Google's Gemini Enterprise app, which was introduced in the fall. Through this app, non-technical users can work with agents built by IT teams or create their own agents to handle tasks such as scheduling meetings, running trigger-based workflows, setting up shortcuts for repetitive actions, and creating or editing files without switching between apps, according to Google.

Google also highlighted that the underlying models powering these tools include its own Gemini large language model and Nano Banana 2 image-generation model, as well as support for Anthropic's Claude. The company confirmed compatibility with Claude Opus, Sonnet, and Haiku, covering flagship, reasoning, and lower-cost model tiers, including the recently announced Opus 4.7.