The evolution of online search gives us a useful prologue to explore why multi-nationals still struggle to go big with AI. Not long ago, most queries were functional and short: “White dress shirt.”
Now the average search contains 15 to 20 words, rich with intent, preference and context: “I want a white shirt, tight fit, button-down collar from store A, B or C. Pickup tomorrow.”
The shift is not because people have more to say. It’s because people increasingly expect systems powered by AI to interpret nuance. When you can loosely describe what you want, and the technology returns something useful, the stakes grow. The minimum expectation becomes that the digital systems we use understand what we need and increasingly, with agentic, can act for us.
Applied to enterprise technology, this shift upsets the traditional balance of power.
For decades, bold moves in enterprise IT happened when leadership was ready.
For decades, bold moves in enterprise IT happened when leadership was ready. Strategy teams defined priorities and transformation offices orchestrated modernization programs, each developing multiyear roadmaps that portrayed sequences of adoption. Employees at the edges of the business, and often closer to customers, had no choice but to follow the plan.
Then AI showed up, arguably before most executives planned for it to, and it undermined the logic. Because AI moves faster than strategy can absorb, the assumptions that once governed enterprise technology no longer hold. Now, thanks to AI in everyone’s pocket and the dynamics of globally competitive markets, AI is a bold move that’s being done to leadership. Less than a third of leaders believe their workforce is ready for AI.1
Execs at HQ are not the only ones feeling it. Well-intentioned implementations of AI can roll out to field teams without adequate training or context or, depending on company culture, credible assurances that employees will be valued after they’ve trained AI to do parts of their job.
Central teams are flummoxed. Workers at the edge balance concerns about their own careers with firsthand experiences of customer expectations, operational friction and real-world constraints. Yet they are exactly the people who can help define AI jobs-to-be-done that will boost productivity, customer satisfaction and other key performance indicators.
We’re moving along the adoption curve to a point where winning with AI will be less about the maturity of the solution than about its widespread adoption. To keep pace, leadership must lean into the idea that the competitive edge from AI will flow from the edges, not the center.
Why centrally led transformation can’t keep pace
Enterprise technology models are built on stability: define the problem, design the system, roll it out. AI upsets the stability in four ways.
Consumer expectations move faster than enterprise systems
Employees use fast, conversational multimodal AI on their own devices. Then they log into systems held together by workflows, tickets and multistep menus. The experience gap becomes a source of friction. Enterprise systems can’t reason, summarize or interpret context the way consumer AI can.
AI evolves faster than planning cycles
New models and architecture emerge every two to three months. That renders outdated any approach to planning adoption over several years, as we did for example with virtualized infrastructure and cloud. Three-year transformation roadmaps can’t keep up with three-month capability cycles. A static architecture approval for three or four years just doesn't work anymore. In the past year alone, enterprise focus has shifted from model selection, to platform consolidation, to deploying agentic workflows that automate complex business processes – a cycle of change measured in quarters, not years, and far outpacing legacy IT planning horizons.2
Micro-frictions are only visible from the edge
Central strategy teams can model processes and design workflows. But what programs are in place for them to see and feel and understand the lived friction of day-to-day work; where consumer expectations meet enterprise limitations and AI solutions may have tremendous value? These frictions are rarely visible to senior leadership or transformation teams. But they are immediately visible to employees interfacing with customers and systems every day.
Demands for data sovereignty create new gates
Across Europe and other regions, organizations now require:
- Nationally isolated endpoints
- In-region data processing
- Residency controls
- Clear policies governing model output
This is not optional and it is clear leaders are increasingly recognizing how critical data sovereignty is. Indeed, 41% of leaders are repatriating at least some data to on-premises environments and geopolitical volatility is only going to push this number up. Sovereignty requirements create a gap between global governance and local acceleration. Sovereignty shapes where AI can run, what data it can access and what agents can do. It shapes what the edge can do. Without sovereign infrastructure, frontline teams cannot experiment safely, even when they have the insight and motivation to improve offerings.3
Thanks to AI in everyone’s pocket and the dynamics of globally competitive markets, AI is a bold move that's being done to leadership.
A different way to think about enterprise AI
Given the pressures, we need to flip the model for technology-led innovation. We need the competitive advantage with AI to grow from the edges inward in the following ways:
Push capability to the edge
Companies don’t have ideas, people do. Companies that are accelerating the fastest right now are those that are giving the tools to their individual users at the edge. That is where brilliance happens. It’s not by businesses that build vast enterprise systems. It’s by giving the tools to everyone at the edge who can help drive greater economic growth because they understand specifically what the customer need is at that point in time. Through agentic systems and safeguards and governing, we can make sure that the right controls are in place centrally to enable innovation to happen locally, without necessarily going back to the CIO or the Chief Technology Officer or the head of business line.
There's always been a gap between someone having an idea and bringing that to life. In the before-agentic era, someone with an idea would file a requirements document with suggested use cases. The idea may eventually have been greenlit, depending on cost and value. But it just as easily may have been stifled in the process.
Now everyone in the organization can have access to governed AI tools that can access enterprise data securely and in line with policy. Employees can now vibe code on a platform that brings the idea or functionality to life and enables them to drive innovation. The best example I have seen of this is between two retailers who decided to automate at the store level, not even at the central level, the ability to take in a stock and inventory sheet and then proactively mail the customers that had asked for products to be reserved for them in store. It was purely as simple as taking one data sheet, merging it with another and finding an outcome. The difference was this was done by two people who had no knowledge of how to do that level of technical integration. It was done through a prompt, which was “Can you help me merge these two sheets and create emails to the people that are on the list?” And they did it themselves with no intervention from IT. Customer service and satisfaction went up and when it was explored as to why, it was found that it was due to this point of innovation that happened purely because the organization had enabled the employees to have the ideas and to bring them to life.
Start with a global governance platform
AI is an enabler, not a brand-new strategy. It always worries me when I meet with a team that presents their “AI strategy.” The leaders’ responsibility is to understand what the corporate strategy or national strategy is, and how AI can be an accelerant or a capability driver within that strategy. And an important aspect of this is to have a global governance platform to ensure regulatory compliance in each region.
Plan for how the company will adopt technology
The most common question that I field from businesses is, “What should I do first?”
The answer is always regulatory and governance: making sure that you can do things safely and in a protected way. Then it’s about the implementation of a knowledge management platform, which looks at how we aggregate every system in an enterprise and make it universally accessible and useful.
The moment to act is now
Now after-agentic, we have reached a point where economic growth based on AI innovation will only start to accelerate. Again: it's not a conversation around maturity, it's a conversation around adoption.
The organizations that will win are those that democratize access to AI for all of their employees. This is not a conversation about productivity. It’s about providing access to data across all enterprise systems, so employees can make informed, structured decisions that drive the business forward in a highly autonomous way while maintaining the organization’s security safeguards.
Employees already know what excellent digital experiences look like. They see where customers struggle. They are ready to act. The question is if their leaders will let them.
Have questions or want to learn more?
