Standing in a colossal building that’s housed multiple engineering marvels, CEO Thomas Kurian put the crowning layer on Google Cloud’s market-leading AI stack with Gemini agent, which Kurian promised would set new standards for delivering business outcomes securely, rapidly, and affordably.
While some attention-seeking “experts” will no doubt snipe that Gemini agent is a me-too response to recent launches of personal AI breakthroughs — particularly Meta’s Muse and OpenAI’s Dots — those doing so will as always miss the real significance. And that is Google Cloud’s creation of a thoroughly modern agent that not only offers personal AI capabilities but also fits elegantly into Google Cloud’s market-leading enterprise-AI portfolio.
During his keynote address at the Gemini At Work event at the iconic Moffett Field Hangar One facility last week, Kurian noted Gemini agent’s ability to handle personal-AI chores but devoted the overwhelming majority of his remarks to its potential as a primary driver of enterprise AI transformation. While you can get the full overview of Google Cloud’s expectations for Gemini agent at the link above, here are some reasons why I believe Google Cloud has dramatically raised the bar for what customers will expect all agentic-AI players to deliver.
1. Three Commitments Competitors Will Have to Match — Or At Least Address.
Clearly acknowledging the intense urgency customers are facing here in the AI Revolution, Kurian made three promises to customers and prospects that go directly to their primary concerns as their companies and industries undergo unprecedented upheaval.
“The hardest part of this is rarely the technology. It is changing how a company works, and that takes longer than any of us would like. As you build on this platform, we make three commitments to you:
- First, we take the friction out by giving you a future-proof platform — runs everywhere, does any task, with your choice of models, and complete controls.
- Second, we eliminate complexity. Google builds and integrates the whole stack — the agent, the tools, the skills, the models, and the controls.
- Third, what you build stays yours. We do not ask your people to move, we do not ask your data to move, and we do not ask you to hand over what you make with it. And we deliver this together with our global partner ecosystem. Inside Gemini Enterprise, you have the broadest choice of third-party agents from leading SaaS providers and connectors available anywhere, ready to use without building them yourself, which you can now discover as collections in the new AI Solutions Finder. Around it, a services ecosystem that is scaling faster than anything I have seen.
2. AI Affordability Is a Function of How Complete a Vendor’s AI Stack Is.
Kurian candidly addressed a jarring statistic reflecting the huge concerns that customers have about the cost of enterprise AI: “Per-token prices have fallen 98% since 2024, and enterprise AI spend is climbing anyway,” Kurian said, “and the reason is volume. A single complex agentic task can consume a hundred thousand tokens.”
In response, Google Cloud has developed a three-pronged approach to help customers manage AI costs: full choice of which models to use, intelligent routing, and automated spending caps per project. From Kurian:
“All of this — the agent, the data, the identity, the cost — rests on two layers most people never see. They are also the reason the economics work. Start with silicon. We design the chip, the interconnect, the data center, and software as one system, delivering up to 80% better inference price-performance, making routine agent tasks cost less than 25% of what they cost elsewhere. On top of the silicon sit the models: Gemini for frontier reasoning, Flash for volume, Omni for generative media, and Gemma, our open models, for the workloads that have to run somewhere else entirely. And I mean anywhere.”
3. The Need for Speed.
In these dizzying times of relentless technological change, Kurian said that customers are more concerned about rapidly delivering outcomes than exploring the technical wizardry of various AI tools. “The question I get most often is not whether this works — it is how long it takes to stand up and get everything working,” Kurian said. He then shared this head-spinning customer story: “Liverpool, one of Mexico’s largest retailers, built a product discovery agent on Gemini Enterprise and had it live in one week. One week, and it is forecast to return 10X what it cost” (emphasis added).
4. Transcending Boundaries Across Vendors, Apps, Solutions, Data Stores — And Say Farewell to Egress Fees!
A key premise of agentic AI is its ability to deliver superior business outcomes by being able to work end to end across an enterprise. And that means being able to tap into and leverage data from a bewildering number of places, which is an idea that until very recently has been rendered impossible by technological obstacles. But Gemini agent overcomes those barriers, Kurian said.
“Your data lives across clouds, in SaaS applications, and on-premises. We are not asking you to move your data. Our agent uses our borderless Lakehouse to query Amazon S3 and Azure Data Lake with no egress fees. It reads Salesforce, SAP, ServiceNow, and Workday without copying anything. And it federates open Iceberg across Databricks Unity, Snowflake Polaris, and AWS Glue.”
5. ‘The Prompt Window Is the New Interface’ — Why Does that Matter?
Kurian didn’t hedge on calling out that massive shift, adding that the prompt window “is not one of several — it is the interface.” And in that context, he said, “the more important question is, ‘What is behind it’?” He answered that by offering six ways in which Gemini agent is differentiated from any other offering:
a) One single agent to engage.
“The same agent answers your questions, does your knowledge work, creates your images and media, and writes and runs your code — all in one window,” Kurian said. “Other tools give you one product for chat, another for tasks, and a third for code. This is one agent doing all of it.”
b) Harmonized memories, context, personalization.
While competitors offer “a desktop agent and a separate cloud agent that do not share memory,” Kurian said, Gemini agent runs in the cloud so comes across exactly the same “on the web, your phone, the desktop, the command line, in Google Workspace, and in Microsoft 365, ServiceNow, and Slack.”
c) Team-builder.
To handle large-scale jobs, Gemini agent “can dynamically create a roster of co-worker agents, assign each of them tasks, message them directly, and coordinate them until the work is completed,” Kurian said. “Each co-worker has its own email address, its own Drive, its own virtual machine, and its own browser.”
d) Full organizational awareness.
Gemini agent understands “your tools, your skills, your data, and your own history of how you work” by tapping into “everywhere the work happens: Workspace, Gemini Enterprise, Slack, and Microsoft Office. It holds this information as a single personalization record, and the longer you work with it, the better it gets to know you.”
e) Complete governance.
From the get-go, Gemini agent inherits and knows all of your enterprise controls.
f) Customers have free rein to use any models they want.
“The agent is separate from the model underneath, which is a choice: Gemini, Claude, GPT, or open models like GLM, whichever fits the job,” Kurian said.
Final Thought
Reinforcing its position as the #1 company on the Cloud Wars Top 10, Google Cloud continues to deliver big, broad, and bold approaches when most other vendors are proposing point-solutions and/or smaller-scale vision.
Kurian, rather than get bogged down in the undeniable technological wizardry that makes Gemini agent possible, instead tied Gemini agent’s broad capabilities directly to the business challenges customers are facing and the business outcomes they’re pursuing.
And that’s exactly why I believe the launch of Gemini Agent — and the potential of the product itself within the Gemini portfolio — will redefine in the minds of customers what agents are and should be.
