The study, which is Google Cloud’s second annual look at what companies are getting out of AI, was conducted by the National Research Group and surveyed 3,466 senior leaders of global businesses about generative AI development in their companies. Oliver Parker, vice president for Global Generative AI Go-To-Market with Google Cloud, said that this year’s advancement in ROI moved the conversation about AI in the enterprise to one of “if” to one of “how fast.” 

“I’ve never seen something move so quickly with so much value,” he told me.

Today’s entire Forbes CIO newsletter is dedicated to this year’s ROI of AI report and my conversation with Parker about what Google Cloud found out about the uses of AI, how it’s impacting businesses around the world, and where things are likely to go next.

If you like what you read here, you can easily share it online and on your social media pages. This newsletter, and all previous editions of Forbes CIO, can be found on our website here.

Parker said that this success changes the way many companies look at AI. Even a year ago, it was considered a separate technology initiative in many enterprises.

“All the feedback coming through the survey represents that this is now truly entering into the enterprise, and it will be woven into the fabric of a lot of core processes over time,” Parker said.

The ROI reaches staggering heights over time. The report found that the average three-year ROI for a business using Google Cloud generative AI is 727%.

Because businesses are seeing AI as a more general technology, they’re also starting to look at new ways to use it throughout departments and functions. And the fact that tech companies are springing up, spreading out and developing new apps to work in this space is also furthering  AI use.

“There’s a level of general industry acceleration, for people to take advantage of these technologies, to continue to be able to add value to their consumer or client experiences,” Parker said. 

Parker said that there are many reasons that companies are using so many agents. Agents can do a wide range of tasks, ranging from taking simple and routine actions to common tasks that once involved coordination of many employees and departments. Nearly half of companies use agents for customer service and experience functions, but they’re also often used for cybersecurity, marketing and research.

Companies generally get started with agents by automating a simple task that is necessary, but considered drudgery for a more skilled employee. Getting agents off the ground quickly can make a big difference to the employee, who can then spend more time on more complex pursuits. That quickly signifies to the company that it’s a worthwhile investment, worth trying for other similarly dull—but needed—tasks. And success in small, less risky tasks breeds enthusiasm for those that are more difficult.

“There’s still a lot of low-hanging fruit, just to be clear, but as those organizations start doing those, I think then they’ll start to get much more sophisticated use cases come up,” Parker said. “Maybe complex system engagement, maybe it could be semi-autonomous. I think over time, you’ll end up with more autonomous workflows.”

While AI in general has a high ROI, those numbers get larger when looking just at agentic AI. When looking at early adopters—defined as companies that said AI agents are deeply embedded across operations and are spending at least 50% of their AI budget on agents—88% are seeing ROI in at least one use case.

“If it’s a process that’s got four steps in it, I’ve just made each step easier. But do I need four steps, or do I even need to recreate a new process?” Parker said, recounting the thought process of business leaders. “This kind of technology, it got people thinking, ‘Okay, is the way that I run compliance and I do these five or six checks. Is that the right way to do it?’”

Figuring out the most efficient way to get business done with AI is certainly being hashed out by leaders, but Parker said they’re also turning to AI for suggestions. It makes sense, he said—executives have seen what a help AI is, and have trained systems on their data, so AI may be able to provide some new insights on process. Parker said this is a “virtuous loop,” with AI helping to constantly improve business processes.

The new processes, Parker said, have also helped some businesses expand what they do, a move that increases revenue and productivity, and helps employees see job security.

Another sign of top-level support is continued investment in AI. As the economy impacts budgets, companies reported spending more on AI, the report states. When other tech costs are falling, 77% of companies are actually spending more on generative AI. More than half (58%) said they are using any new revenues to invest in AI, while 48% are reallocating existing budgets to AI improvements.

“This technology is starting to separate those that have adopted it, and adopted it in the right way, [to] become leaders in industry, whereas maybe they were a leader before but not as a clear leader or maybe it wasn’t really clear who were the leaders in industry,” Parker said. “You’ll start to see the adoption of this technology leading to what I would say is the truest measure of ROI, which is accelerating industry capability and leaving others behind.”

The way that companies are evaluating AI also shows increasing maturity in enterprise use of the technology, Parker said. Data security and privacy is now the top factor companies consider when considering LLM providers—37% prioritize security, 10% more than those placing the same priority on cost. Parker said this is showing the beginning of a larger market shift: Companies see AI as central to the organization and no longer as an exciting and new thing to add because it’s trending.

Another evolution Parker says is likely to continue is the shifting employee perception from AI as something that is brought in to take their jobs away to a tool that helps them do their work better. In his conversations with companies, Parker said he’s seen fewer people fear that AI will replace them—and especially at places where AI is used on a daily basis and in a meaningful way.

“That makes a material difference,” Parker said. “I think you’ll see the adaptation of knowledge workers and employees seeing this as a high-value accelerator to their careers and their capability, versus a diminisher.” 

There are so many decisions business leaders need to make nowadays, and decision fatigue can sap your mental energy. Here are some ChatGPT prompts to help you figure out how to minimize decision fatigue and keep your day going without extra interruptions.