Locked in a high-stakes AI race with Anthropic, OpenAI and its fellow tech giants, Google sought to gain an advantage in the agentic arena, releasing a slate of tools for enterprises to build, manage, scale and secure AI agents.
On Wednesday, the first day of its Cloud Next ’26 conference in Las Vegas, Google introduced Gemini Enterprise Agent Platform, Agentic Data Cloud, Gemini Enterprise App and other systems that reveal its strategy for agentic development and ways it is seeking to match, exceed or catch up to its competitors.
The most important of the new platforms, Google executives said, is Gemini Enterprise Agent Platform, which blends the model building and tuning services of the vendor’s popular Vertex AI platform with new features for agent integration, security, DevOps and orchestration.
“The platform … gives you the full set of key tools to build and deploy agents,” Thomas Kurian, CEO of Google Cloud, said during a media briefing.
Users of the platform have access to generative AI models such as Gemini 3.1 Pro, Gemini 3.1 Flash Image — also known as Nano Banana 2, and Lyria 3, Google’s model for generating audio and music. Customers can also use Anthropic models such as Claude Opus, Sonnet and Haiku and get support for Claude Opus 4.7.
Users can build agents using tools such as the Agent Developer Kit, Agent Studio, Agent Registry and the Model Context Protocol. They can also scale and orchestrate agents with a set of services highlighted by agent-to-agent orchestration, which enables agents to delegate tasks to each other, and runtime improvements.
The Platform
For Google, the Gemini Enterprise Agent Platform helps organizations address the biggest challenges they face when using agents. Similar to OpenAI and Anthropic, Google is working to make Gemini more than just a model and into a platform on which enterprises can build and innovate on.
“It’s much more than just a model as a platform,“ said Bradley Shimmin, an analyst at Futurum Group. “It is the entirety of Google’s cloud stack that they’re trying to bring to bear.“
The platform focuses on a problem that emerged for enterprises, being able to “contextualize AI agents for specific organizations and individuals,” said Ed Anderson, an analyst at Gartner.
“Gemini Enterprise Agent Platform brings together all the elements an individual needs to build and operate AI agents specific to their role and environment,” Anderson said.
However, Shimmin said the Gemini Enterprise Agent platform also appears to be more of an evolution of what previously existed in Vertex AI, which was introduced as a machine learning platform in 2021, six months before OpenAI’s groundbreaking release of ChatGPT and the start of the generative AI era. Vertex AI has turned into Google’s managed platform for building and deploying generative AI applications.
Agentic Data Cloud
Shimmin said that, among Google’s new offerings introduced at the conference so far, what stands out most to him is the Agentic Data Cloud, a new data architecture.
The data architecture features a cross-cloud lakehouse that enables enterprises to keep their data in AWS or Microsoft Azure without being locked into a specific vendor. It also includes a Deep Research Agent for enterprise insight, which combines research and analytical capabilities. This enables the Gemini Deep Research agent to connect to data platforms such as Google’s long-established BigQuery enterprise data warehouse, enabling enterprises to get a full picture of their structured and unstructured data, Google said.
The Agentic Data Cloud shows how Google’s data platform is catching up with its models, Shimmin said.
“What it really does is bring them back into play against their rivals and partners,” he said. “This is the current pattern that we’re really seeing in the enterprise for bringing data to models … enabling agentic solutions that aren’t just hallucinating because of gaps in their context,” Shimmin continued. He added that Agentic Data Cloud enables Google to align its data platform with current agentic practices.
Gemini Enterprise App
Google also introduced new capabilities in the Gemini Enterprise app that help enterprise employees build with AI, such as an Agent Designer that can create sophisticated schedules; long-running agents that execute complex business processes; an inbox for managing agent activity; and skills for creating agents for repetitive tasks.
“Gemini Enterprise App is an incredibly powerful tool to help individuals and teams use AI to interact with their enterprise environment,” Anderson said. He added that combined with the Agentic Data Cloud, the Gemini models and the Gemini Enterprise Agent Platform, the app helps enterprises “gain the productivity and efficiency outcomes they’ve been seeking from their AI solutions.”
“Gemini Enterprise takes a big step forward in making generative AI accessible and usable, and applicable to the tasks people work on every day,” Anderson continued. “The next step will be to apply agentic AI capabilities to the scenarios that produce meaningful business outcomes.”
While the new capabilities in the Gemini Enterprise App may suggest that Google is competing with Anthropic Claude Cowork and other similar tools, Google’s entry into this space bolsters the overall AI market.
“This market thrives on competition in the form of innovation,”” Shimmin said.
Cybersecurity Moves
While building an agentic AI infrastructure is an important step for Google, the vendor is also looking to help enterprises secure the AI agents and applications that they build, with AI cybersecurity top of mind due to Anthropic’s limited release of its Mythos security model, which has caused widespread concern about the technology falling into the hands of bad actors to crack software defenses easily.
Meanwhile, Google introduced new tools for agentic defense, such as Dark Web Intelligence, which uses the newest Gemini models to analyze daily external events and elevate the threats that matter most to an organization. The Threat Hunting Agent uses Google’s threat intelligence technology to enable teams to hunt for novel attack patterns and adversary behaviors that bypass other types of defenses. Google also unveiled Google Cloud Fraud Defense, a platform that can determine whether a bot, a human or an agent has the necessary authority.
On the workspace front, Google introduced the Workspace Intelligence platform, a semantic layer that breaks down information and context for enterprises and their agents. It works across Google Workspace apps and powers AI Overviews in Gmail, Gemini in Google Chat and Content Creation in Docs, Sheets and Slides.
