Are you worried? Think you’re not getting enough ROI from AI? If so, you’re in good company. Enterprise adoption of AI is growing, with one recent report showing that 88% of organizations are using AI in at least one function, but many companies are realizing only modest benefits or, in a small number of cases, no measurable value at all. For leaders, the risks are real. In a global poll conducted in 2025, three out of four CEOs said they were worried they would lose their jobs within two years if the situation didn’t change.
Of course, the corporate hierarchy teeth We believe there are real benefits to be gained from AI. Looking at the finer details will give you clues as to what is being done correctly. Research shows that organizations that fundamentally redesign their workflows to account for AI capabilities are more likely to leverage its full potential. This makes sense, as it reflects how general-purpose technologies have entered the economy in the past. For example, electricity made its presence known within industry, not just as a way to electrically light traditional factories, but by enabling entirely new manufacturing approaches.
The shift from treating AI as a bolt-on technology (overlaying existing processes) to using it as the key to unlocking new ways of working has a name: “AI native.” But for companies not born into the AI era, embracing that ethos involves a dramatic shift in mindset, often requiring companies to think about things they don’t even know they need to think about.
Here, three leading practitioners provide key insights for corporate leaders on that journey. Learn more about their thoughts and others on these topics in a five-part digital event on how to adapt, compete, and win in the AI-native business era.
Be comfortable with radical uncertainty
Benedict Evans is one of the world’s most respected technology analysts, helping leaders understand what matters most in a rapidly changing digital environment.
“Generative AI needs to anticipate changes on the same scale as what happened with the internet and mobile,” says Benedict Evans. This suggests that the most successful companies in the AI era will not only bring new efficiencies to established business practices, but also invent entirely new kinds of AI-native products, services, and revenue strategies. Just like the web gave us social media and search giants. Some of these new AI businesses, like Uber and Airbnb, may turn existing categories inside out.
The challenge for leaders is that identifying these profitable new opportunities is inherently difficult because it’s hard to imagine doing something that hasn’t been done before. The good news is that previous platform migrations have provided us with some mental models to help ensure disciplined thinking.
The first is: What happens when something expensive becomes significantly cheaper or free? When a new, general-purpose technology dramatically reduces the cost of something, it tends to unleash even more profound disruption. “If you wanted to run an express train from London to Edinburgh in 1750, it didn’t matter how much money you had. You could have bought 20,000 horses and put them at the front, but that still wouldn’t work. Then the steam engine came along and you could do it,” Evans says. “So what will AI make cheaper or free? And what will it change? Will it allow us to compete with people we couldn’t compete with before? What new things will we be able to do now?”
The second framework is to consider the types of questions that computers have not been able to address before. I’m going to buy clothes. “The old question is, ‘Here’s a coat. Where can I buy it?’ Now you can use Google. The new question is, ‘Suggest 10 or 15 coats like this at various prices.’ And the next question might be something like, ‘Look at my Instagram and give me some suggestions for coats that will refresh my look and match what I already like.’
Evans’ final tip is to not dismiss counterintuitive ideas out of thin air. “You can’t look at something and say, ‘That won’t work.’ You have to adopt the classic venture capital framework of ‘Will it work?’ ” He points out that most of the things we do with our phones, the idea of even owning one would not have been a given 25 years ago. “New things that are very important always look kind of stupid.”
The trouble with all of this is that we still don’t know the physical constraints on what is possible with AI. From a functionality perspective, unlike previous technologies such as mobile, this is a “moving target”. Therefore, this is a time when strategic inquiry must continue to evolve as developments occur. “Assume fundamental uncertainty: Everything changes, but you don’t really know how,” Evans says. “You always have to stay curious.”
An AI-native business needs an AI-native IT strategy
Michael Bradshaw is a global practice leader for applications, data, and AI and former CIO of Kyndryl, a leading provider of mission-critical enterprise technology services. The company helps global companies design, build, and operate technical and human systems to extend AI capabilities and drive business value.
Nowadays, corporate technology strategies are teeth its business strategy. Technology defines how companies operate and what outcomes they can achieve. But there is a fundamental friction. Agent AI presents radical new opportunities, but most enterprise IT was built in a different era.
In time, organizations have designed IT to enable optimization of individual business functions. As a result, data and applications are often siled across the enterprise. AI-native success requires a different approach, as AI’s greatest potential lies in its ability to deliver enterprise value across functions and at scale. “It’s more about changing the way we do business across the board, from customer engagement to delivery, and using that as a differentiator to drive new dynamics in the market.”
In Michael Bradshaw’s view, that can be difficult to achieve as many organizations’ IT assets have grown in complexity over the decades. As business needs evolve, companies are layering new systems on top of old systems rather than replacing them, creating what Bradshaw calls “scaffolding” – rigidly coupled environments that are difficult to update.
Bradshaw’s advice is that you don’t have to scrap everything and start over. Instead, organizations can work systematically across their assets to determine what to keep, replace, or reuse. To become AI native, companies must direct this effort toward creating a unified data platform that runs “horizontally” across the business. They also need to establish modular, configurable architectures that can adapt to technology advances, rather than relying on siled, vendor-bound systems.
Agentic AI itself helps enable a horizontal and more agile transition to IT. Agents can roam on the backend across The system accesses data directly, regardless of where it resides, rather than being locked to a predefined application. Agents can also provide great flexibility to end users. Current work spans multiple SaaS (Software as a Service) applications that are struggling to meet changing business needs. “In the old model, we had software customized for us, but by the time a developer delivered custom software, the opportunity had been lost or the business had changed.” Organizations can now design workflows in real time to fit the way work is done, making them even faster and more adaptable.
But simply transforming the system is not enough. For Bradshaw, the real challenges to becoming an AI native are not just technical, but also cultural. “This is a people problem, not a technology problem,” he says. “Even though they may have access to great technology, organizations often struggle to bring about change in their employees.” But change is the point. Becoming AI-native means companies move beyond siled, function-driven ways of working to align around shared outcomes with teams working across the business. “It’s a completely different way of thinking about IT.”
Leadership is the most important technical input
Kathy Kozyrkov, Google’s first chief decision scientist and founder of the field of decision intelligence, is a pioneering thinker on AI-human collaboration and the pivotal role of leadership in unlocking the true power of enterprise AI systems.
Kathy Kozyrkov often sees business leaders making categorization mistakes. They treat AI as if it were an autonomous, intelligent system, leading them to believe that human decision-making is less important. “But decisions matter more than leaders realize,” Kozyrkov said.
As an analogy, she mentions the metaphor of the genie in the lamp. Genies have great powers, but they can sometimes misunderstand wishes and act in unexpected ways. The same applies to AI, which inherently doesn’t know what’s right or wrong. Achieving desired results depends on how you design your prompts, policies, guardrails, and harnesses, and whether you intentionally review results on a regular basis to avoid agent drift.
While it may sound like a simple statement of fact, she is at pains to highlight how companies overlook the need to be “decision-first” at both a technical and strategic level. Before you build, it’s important to know exactly what you want to achieve, what level of risk you’re willing to take, and how you intend to measure it. You can’t just hand out tools and let people do whatever they want with them.
Leadership therefore becomes a “technical input” to generative AI systems, she says. Without it, the project will fail.
But it’s not easy to know exactly what we want to achieve from the technology, which is still being developed at a breakneck pace.
One effective tactic is to make a list of things you’d like to do with AI to transform your business once the technology is mature enough, and return to the list periodically. “That could be anything from an increased proportion of a process to a completely different product workflow,” she says. “Whatever it is, write it down. Relatively soon, AI may be able to provide it to you. Every six months. minimumgo back to your list and see what’s possible that wasn’t possible before. ”
The nature of AI-related decision-making also places a new emphasis on specific skills. To unlock the true potential of generative AI, we need to think about it not as a standalone technology, but as part of a broader architecture, both in terms of opportunities and risks. This dual requirement means that “systems thinkers” – people who can naturally intuit holistic relationships, feedback loops, and patterns – are especially valuable. “I’ve seen a lot of leaders who don’t realize that there are actually a lot of untapped systems thinkers in their organizations,” she says. “My favorite way to tell them apart is to ask: ‘Who do people tend to turn to when something goes wrong? Who is the person who finds fault?’
Often, leaders will instead look for people who may have specific degrees or certifications to help them with their AI efforts, she says. “But what is the context behind actually preparing for this radical and brave new world of AI?”
These are just some of the things Ben, Michael, and Kathy will cover in their June 5th live broadcast, “Rethinking AI: Becoming an AI Native.” Join them and other leading experts in AI and business to hear bold perspectives, practical insights, and real-world case studies to help you become an AI native. Sign up for free here for a chance to ask our experts your burning questions.
