Who Owns AI Transformation?
Observations from the UNLEASH America Conference
Last week, I found myself in a room full of people I don’t usually hang out with professionally.
Thanks to Stacia Garr and the Red Thread Research team, I found myself at UNLEASH America, an event focused on reinventing work. UNLEASH put me squarely in the people side: CHROs, heads of talent, L&D leaders, HR technology practitioners. Brilliant people wrestling with questions I typically address from a different vantage point.
And what I found there clarified something I’ve been circling for months: Who owns AI transformation? Truly integrating AI is a whole-enterprise problem. And right now, nobody quite owns it. The ambiguity is stalling organizations out.
A State of Confusion on Who Is Driving What
Analyst Josh Bersin provided a keynote that was equal parts data deluge and honest editorial (his words, not mine). He’s been traveling the world meeting CHROs and HR leaders, and his diagnosis of the moment didn’t hold back: confusion. Not ignorance. Not reluctance. Confusion. The kind that signals you’re standing in the middle of an epic shift from one state of the world to another, and it is unclear who should be in the driver’s seat.
IT spending on AI is up nearly 62%. Does that mean IT should own the transition? Some 40% of companies are now spending $10 million a year or more on AI. Some of this is training. Should HR be driving the bus? The average large enterprise runs more than 1,000 distinct technology systems (!), 97 of which are employee-facing, almost all of them integrating AI into their software. Should the vendors supply the knowledge?
In part, because of this confusion, most organizations are moving at a fraction of the speed the technology, Bersin noted. AI capability is advancing on a steep upward curve. Organizations, with their jobs, structures, cultures, legacy systems, and deeply human resistance to change, are moving on a much shallower one.
My take is that we haven’t figured out we need to ‘invest in the rest’ when it comes to AI. For years, I’ve said shopping is the easy part. Installation is where the friction appears. And when you look at the above list — jobs, structures, legacy systems, human resistance — it takes an deliberate amount of time and resources to rewire these parts of the business. We know from the past that ignoring the people part of technology leads to delays and frustration. AI is no different.
Where AI Diverges the Past
Everyone I talk to (or hear speak), from the front lines of those building the models to the small business owner is breathless is overwhelmed by the non-stop flood of features, advancements, and hype around AI
Previous waves followed predictable adoption life cycles, years long, with room to plan, pilot, and scale. AI is doing none of that. Model updates are dropping every six weeks. Employees are already using tools their employers haven’t sanctioned. Your competitor isn’t waiting for you to complete your governance review.
So, how do we keep up with it all? Bersin outlined a deliberate progression (one that maps remarkably well to the Hyperadaptive journey; see below).
He said organizations start with individuals using AI to do their current jobs better, writing faster, analyzing data more quickly, reading emails with more efficiency. That’s real, but it’s table stakes. The next move is automating repeatable tasks, building reusable workflows. Then comes the moment where those individual automations start talking to each other, multi-agent coordination. And finally, entire business processes get reimagined around super-agents that orchestrate end-to-end work no single system ever touched before.
His example: UKG has built a super-agent capable of handling a front-line worker’s request, such as “I need an extra $400 after tax before Christmas, can you find me the right shifts?” The agent simultaneously checks qualifications, identifies eligible shifts, optimizes for pay rate, calculates tax impact, and returns a complete recommendation. That’s the future. Coordinated intelligence operating across what used to be five separate tools and three human handoffs.
Addressing Real Fears Around AI
The people side of the house isn’t just confused about technology. They’re watching their workforce become frightened and trying to figure out what to do about it.
Bersin cited University of Michigan consumer confidence data showing that American workers are experiencing a 50-year low in how comfortable they feel about their economic futures. He was careful to note that this isn’t just about inflation or income inequality. There is a palpable fear factor layered on top. People are reading the headlines about jobs being eliminated by AI, and many of them don’t have an understanding what happens to them in that story.
For large swaths of the workforce, reinventing themselves professionally is terrifying. They don’t know how. They don’t know where to turn for support. They don’t have models for it. The AI video library isn’t cutting it. And they’re watching headlines that suggest they may have to, whether they want to or not.
This matters because the most sophisticated agent architecture in the world fails if the humans it’s meant to augment don’t trust it, don’t understand it, or have quietly checked out. We need to provide the missing infrastructure.
So, Who Owns This?
In the room at UNLEASH, it was clear that HR leaders see AI transformation as, at least partly, their problem. The organizational change, the reskilling, the talent architecture, the workforce sentiment. All of that is undeniably their domain. But many of them are waiting for IT to lead on the technology side. And many IT leaders are pushing back by pointing to the business process owners. And the business process owners are looking at the C-suite. And the C-suite is making announcements.
Everyone sees the problem. Nobody fully owns it.
I believe this isn’t an HR problem or an IT problem. It’s a systems problem. And systems problems require systems thinking, which doesn’t stop at functional boundaries. Frankly, we need a cross functional AI Transformation office, alongside dedicated support structures that allow the organization to continue to reinvent themselves (I outline suggested ones in the Hyperadaptive Model).
What Bersin described as “systemic HR, stitching together the fragmented specialties of the HR function around actual business problems rather than internal organizational charts,“ is actually a version of something every function needs to do right now.
The siloed approach to AI, where each function pilots its own tools and builds its own workflows and reports its own wins, is how you end up with 97 employee-facing applications that can’t talk to each other and a workforce that’s more confused than empowered.
I created the Hyperadaptive Model because I saw the limits of previous technology transformations. I saw them stall as IT transformed only to hit the wall of finance or HR. I envisioned a lightweight model that cut across functions, giving everyone a shared destination with enough wiggle room to move at their own pace. A model that could stay durable as people moved from individual AI augmentation through process optimization, early automation, scaled AI, to what I call the Hyperadaptive state: an organization that doesn’t just use AI but continuously evolves with it. The model requires governance structures that cut across functions (AI Councils). It requires human nodes in every part of the organization who can translate AI’s potential into local context (AI Leads). It requires the social infrastructure to spread learning faster than any formal training program can (Communities of Practice and AI Activation Hubs).
The blueprint, grounded in research extended for the age of AI and examples from leading company is outlined in the book Hyperadaptive. What we need to do now, is organize around the blueprint and fund it.
What You Can Do With This
If you’re reading this as an AI transformation leader, I’d offer you two challenges coming out of this conference.
First, go find your counterparts in functional areas. Not to hand them the problem, but to build a shared map of it. The architecture questions and the workforce questions are not separable. The leader who understands both will be the one who actually moves the needle.
Second, create your shared roadmap. The research is here. The models exist (I’m sure there are many). What’s missing in most organizations is the organizational will to treat AI transformation as a whole-enterprise priority that extends beyond the technology.
The gap between where the technology is going and where most organizations are standing is real. But it’s not inevitable. It’s a design problem. And design problems have solutions, if you can work together to find them.
Hyperadaptive: Rewiring the Enterprise to Become AI-Native releases May 12th and is available for pre-order now at hyperadaptive.solutions/book. If you order now, you get early access to the model and supporting materials. It is research-grounded, enterprise-focused, and built for exactly the moment we’re all standing in.
And if you’d rather explore these concepts live, I’m hosting a peer roundtable on April 16th, just two weeks away, where we’ll dig into organizational alignment as the AI transformation function nobody assigned. We have a nice group formed. Upgrade to paid to be part of it.