"The era of the pilot is over," Google Cloud CEO Thomas Kurian declared at the company's annual Next conference earlier this mont, "The era of the agent is here."
With more than 32,000 attendees on the ground and over 260 announcements made, spanning infrastructure, data, security and applications, the event marked one of the company's biggest moments yet.
The message was clear: agentic AI is here, and Google Cloud wants to offer solutions across the whole stack.
What agentic AI means for Google
Central to this year's Next's narrative was a shift in how Google envisions organizations interacting with AI as they shift from generative to agentic AI.
For the past two years, Google Cloud explained, most of the industry has focused on passive, copilot-style assistants to help employees write, code or analyze faster.
This year, AI is moving beyond assistance to become an active participant in business operations, often totally separated from a human worker. Agents are being positioned as "intellectual peers" – not just tools – capable of reasoning, coordinating and executing tasks.
Speaking in a pre-recorded message at the event, Google CEO Sundar Pichai revealed that around 75% of the company's new code is AI-generated, but we also now know that around 75% of Google Cloud's customers are now using AI in some capacity.
To support growing interest, the company outlined what it's calling its agentic enterprise blueprint, which rethinks the full stack of enterprise IT architecture. Hence the seemingly endless new product announcements and updates.
What agentic AI means for enterprises
When he stated that the era of the pilot is over, Kurian means to say that investing in a unified platform of data, models, infrastructure and applications is key, rather than layering AI on top of existing, legacy systems.
Speaking to a group of journalists, Agentic Data Cloud VP and GM Andi Gutmans explained that "humans can only click so many times a day." With agents effectively removing that ceiling, operating around the clock and at scale, enterprises must totally rethink their data foundations to maximize AI ROI.
Gutmans stressed that, even today, around 90% of all enterprise data is unstructured 'dark' data that's historically been underused.
Agentic technologies promise automated date discovery, interpretation and enrichment, and Google Cloud is not only at the forefront of this, but it's making strides to meet customers where they are.
The company explicitly acknowledged that enterprises run multi-cloud and hybrid environments beyond just Google Cloud, and it has a clear vision to activate all enterprise data wherever it is, not to "move everything to Google Cloud."
At the same time, company leaders acknowledged that this transition introduces new complexities such as agent management, governance, security and cost, all while systems become highly interconnected and interdependent. Something it aims to tackle with the new Gemini Enterprise platform, which we've covered separately.
At the end of the day, agentic may be here but companies may not be agentic just yet, and Kurian's message serves as a warning message for them to reconsider their stack from the ground up rather than continuing to build and add layers of complexity on top of it.
