Building the “how” in AI transformation
|  | | | |
ON REWIRING ORGANIZATIONS
Timely and timeless in the age of AI
| | | | | | | | | | | |
| Cowriting the second edition of Rewired: How Leading Companies Win with Technology and AI (Wiley, April 2026) was a strange experience.
The technology landscape had changed so much since the first edition. In just a few years, we moved from machine learning models that generated insights for humans to act on to generative AI copilots and now to early agentic systems that can operate autonomously across entire workflows. The acceleration has been radical and the opportunities unprecedented. We are living through a generational moment.
Amid all the change and growing pains, however, the central lesson of the book’s first edition still holds: Turning technology into value depends on building the organizational capabilities to properly harness the tech. We call it the “how engine.” The technology may evolve (from digital to AI to quantum) but the need for a high-performing “how engine” does not.
The importance of this “how” is particularly germane now with AI value realization only just beginning to scale. The latest data from McKinsey’s State of AI report shows that nearly nine in ten organizations are already using AI in at least one function, and many are experimenting with agentic systems. But most companies have not yet begun scaling these capabilities enterprise-wide, and even fewer have deployed agentic AI meaningfully even within a single function.
That gap between technological potential and organizational readiness is what we think of as the AI value paradox: The capability of AI has outpaced companies’ ability to reorganize around it. That’s not unusual. Every major technology wave has exhibited this dynamic. What’s different this time is the pace.
Many of the companies that have been keeping up and thriving as this phase unfolds are the ones that have built strong foundational capabilities. The companies featured in the book’s first edition—Freeport and DBS—were succeeding then and are succeeding now. Not because they picked the right model or vendor, but because they built the capabilities to identify value, redesign work, reskill talent, and scale change across the enterprise while investing in enabling technology and data stack modernization. (This is also true of the other 20 leading companies we analyzed for the book.)
| | |
| |
| “The gap between technological potential and organizational readiness is what we think of as the AI value paradox.” | | | |
|
| Within each of those capability areas, however, AI has catalyzed some significant changes. There’s a lot to say on that front, but here are a few new developments across each of the six capabilities that have particularly struck me:
| | | | | Strategy has always begun with identifying where to unlock value, but with agentic AI that comes from reimagining workflows end to end in targeted domains. This end-to-end practice had the greatest correlation with bottom-line value in our latest State of AI results. | | | | | | | For years, the focus was on hiring scarce AI specialists. That still matters. But the larger shift lies in enabling the entire workforce to work effectively with AI. In addition to finding the “10x technologists,” we now need to consider how to reskill an entire workforce to become 10xers in their areas of expertise. The paradigm of what an AI-enabled worker can do is quickly evolving. | | | | | | | That shift has implications for the operating model. In earlier digital transformations, organizations searched for a handful of exceptional product owners to lead efforts to build and adopt. We would argue this is now broadly the paradigm for every manager of the future. They need to be technology-capable leaders who can manage small, fast-moving teams that combine business, engineering, and AI capabilities. | | | | | | | That shift is most profound in software development, where resourcing and economics are being disrupted. For a new product build, a 12-person development squad can now be replaced by a product owner and a full-stack engineer. Clarity of vision becomes most important, and speed is strategy. This evolution in the ways of working in software development offers a sneak peek at what change could look like for all knowledge work. | | | | | | | Agentic AI has made technology architecture a strategic topic. With AI capabilities now embedded across nearly every layer of the stack, technology leaders must make deliberate choices about where to standardize, where to remain flexible, and where to draw the line on vendor dependency—all while the underlying technology continues to advance faster than most enterprise planning cycles can accommodate. Getting the architecture right is no longer an IT decision; it is a business decision. | | | | | | | Data remains foundational, but its role, too, is evolving. Clean data and well-governed data products are no longer sufficient on their own. Agentic systems need a way to understand relationships across enterprise information. That’s where semantic layers and knowledge graphs come into play, enabling agents to navigate data more intelligently across domains to minimize the time and cost of data integration. | | | | | | | Finally, adoption and scaling may be the capability undergoing the most profound changes because of how broad the AI revolution is. The process of reimagining a full process with AI, redrafting human roles for new tasks and skills, implementing rigorous training, and tracking adoption both internally and with customers, suppliers, and other partners has become a new enterprise core competency. Organizations will need to master the hard work of transformation and repeat it across every domain. That’s something no one has done before. | | | | | Agentic systems will become more capable. The tools will keep improving. But the companies that benefit most will be the ones that can build the “how” to match the “wow.”
| | | —Edited by Barr Seitz, editorial director, New York | | |
| Share Robert Levin’s insights | | | |
|
| |
| | Robert Levin is a senior partner in McKinsey’s Boston office. | | |
| | | | | | |
|
|
|
Copyright © 2026 | McKinsey & Company, 3 World Trade Center, 175 Greenwich Street, New York, NY 10007
|
|
| |
|
|
