AI adoption ownership should be split between IT and HR: IT owns tool selection, security, and integration; HR owns behavior change, manager enablement, and whether people actually use the tool differently. In most companies the whole thing gets handed to IT, and that is exactly why adoption stalls. The human half has no owner.
IT owns the technology. HR owns whether people actually change how they work. Those are two different jobs, and treating them as one is the most common mistake Boon sees across its client base.
The question sounds like a turf war. It isn't. It's a question about where the work actually lives, and the honest answer is that the work is split, and one half of it is currently going undone.
Why IT Ends Up Owning AI Adoption by Default
It makes sense on paper. AI tools are software, and software is IT's domain. So when the CEO says "we need an AI strategy," the ask lands on the CIO's desk, and the CIO does what CIOs do well. They evaluate vendors, run a pilot, secure the data, and ship the thing.
None of that is wrong. It's necessary. You cannot adopt a tool you haven't bought and secured.
But buying and securing a tool is not the same as getting people to use it differently. IT can put Copilot on every desktop in the building. IT cannot make a senior account manager who has done her job the same way for twelve years trust a tool to draft her client emails.
That second problem is a human one. It's about habit, fear, identity, and how someone's role might change. IT is not built to solve it, and most IT leaders will tell you so if you ask them directly. Boon wrote about this pattern in detail in why IT-led AI rollouts stall. The short version: the tool gets deployed, the dashboard shows licenses assigned, and usage flatlines six weeks later because nobody owned the part where people change.
What "Adoption" Actually Means
There's a quiet sloppiness in most AI strategies. "Adoption" gets used to mean "deployment." They are not the same thing.
Deployment is: the tool is live and accessible. Adoption is: people use it to do real work, regularly, in ways that change the output.
You can fully deploy a tool to everyone and still see almost no one actually using it. Boon sees this constantly. The license report looks great. The actual behavior hasn't moved.
The reason is that adoption is a behavior change problem wearing a technology costume. The blockers are almost never technical. They are things like: "I don't trust the output." "I'm worried this is how they replace me." "I tried it once, it was wrong, I went back to my old way." "My manager still wants it done the old way, so why bother."
Research on AI rollouts consistently points at trust, ethics, and fear as the things that stall generative AI more than any feature gap. That tracks with what Boon sees across its client base: the wall isn't capability, it's the people side. None of those blockers have an IT fix. They have a coaching and management fix. There's a good breakdown of the resistance side in how to overcome employee resistance to AI.
Who Owns AI Adoption: A Clean Split
Stop arguing about whether IT or HR owns it. They both own different halves. Here's the split that works, based on what Boon sees in rollouts that succeed:
- Tool selection and security. IT. Full stop. This is their expertise and it should stay there.
- Integration and access. IT. Getting the tool into the workflow, into the systems people already use.
- Policy and governance. Shared. Legal, IT, and HR. What's allowed, what data goes where, what the guardrails are.
- Behavior change and adoption. HR and L&D. This is the part that's currently orphaned.
- Manager enablement. HR. Managers decide whether their teams actually use the tool. Most companies forget this entirely.
- Measuring real usage and capability. Shared, but HR owns the human-outcome side.
Look at that list. Four of the six items touch HR, and the two most important for adoption, items four and five, sit squarely with HR. Yet in most rollouts HR wasn't even in the room when the tool was chosen.
That's the gap. Not a budget gap. An ownership gap.
This Is HR's Moment, and Most Teams Are Missing It
The polite version of this point gets ignored, so Boon will be direct: AI adoption is the clearest business priority in a decade where the technology is the easy part and the people part decides everything. HR has spent years asking to be part of strategy. This is it. In Boon's own client base, the pattern is consistent: the work keeps defaulting to IT, and HR lets it.
Some of that is habit. Some of it is that HR doesn't feel technical enough to claim the territory. But owning this work doesn't mean understanding the model architecture. It means owning the thing HR already understands better than anyone: how people change how they work, and what gets in the way.
If your L&D team can run an onboarding program, they can run an AI adoption program. The muscle is the same. Boon made a version of this argument in how to build an AI adoption strategy: adoption is a change program, not a software launch. HR runs change programs. That's the job.
What HR has to stop doing is waiting to be invited. By the time the tool is deployed and adoption is flat, the moment to shape it has passed. Get in the room at tool selection. Ask one question: "What's our plan for the people who won't want to use this?" If the answer is a help-desk ticket queue, you've found your job.
The Manager Layer Is Where Adoption Lives or Dies
Skip the manager layer and you skip the part that actually decides the outcome, which is exactly what most AI strategies do.
People don't adopt new tools because IT deployed them or because HR ran a webinar. They adopt them because their manager expects it, models it, and makes it safe to fumble through the learning curve.
A manager who says "we're an AI team now, here's how I'm using it, here's where I got it wrong this week" creates adoption. A manager who says nothing, or quietly keeps doing it the old way, kills it. No license report captures this, but it's the single biggest variable Boon sees.
This is why the "train everyone on AI" approach underperforms. A two-hour session teaches features. It does not change what a manager expects on Monday morning. If the manager's behavior doesn't shift, the team's behavior reverts, every time.
Boon has written a lot about why the manager layer carries this kind of weight, in why your managers are the real engine of growth and in the leadership ripple effect. The mechanics are identical for AI. You can't change a hundred people's behavior by talking to a hundred people. You change it by working with their managers, because their team's habits follow how the manager works day to day.
So the real ownership answer gets sharper: HR owns adoption, and HR delivers it through managers. Not through a portal. Through the people those hundred employees actually listen to.
Why "Train Your People on AI" Doesn't Change Behavior
The default HR response to AI is training. Workshops, certifications, a learning library full of prompt guides. It feels like ownership. It mostly isn't.
Training delivers information. Adoption requires behavior change, and those are different problems with different solutions. You can know exactly how to use a tool and still not use it, because the barrier was never knowledge. It was fear, habit, or a lack of permission.
Coaching is what closes that gap, because it works on the actual blocker. A coach doesn't re-explain the feature. A coach works with the person on what's stopping them from using it, which is usually some version of "what does this mean for my job." That's a conversation, not a curriculum.
This is the thinking behind AI transformation coaching: you pair the tool rollout with coaching at the people level, so the change actually lands. Boon's program data shows coaching engagements produce a 23% average improvement in the competencies people are working on, and session attendance runs at 89%. That second number matters more than it looks, because adoption dies in the gap between "assigned" and "actually showed up." People show up for coaching in a way they don't show up for another optional webinar. You can see how the always-on version works on the Boon Scale page.
How to Set Up Shared Ownership That Works
Setting this up isn't complicated. It just requires HR to claim its half on purpose instead of by accident.
Start with a joint owner model where IT and HR are both named, in writing, on the AI adoption effort. Not "IT leads, HR supports." Co-owners, with the split above made explicit. In practice, that looks like a single document that lists who owns tool selection, who owns manager enablement, and who owns the usage metrics, with names next to each line.
Get HR into tool selection early, so the people who'll own adoption have a say in what they're being handed. A tool chosen purely on features, with no thought to how people will fold it into their day, is harder to adopt no matter how good the change program is.
Then build the adoption plan around managers, not around end users directly. Equip managers to model the behavior, set the expectation, and coach their people through the awkward early phase. If you want a sense of how to do that at scale without exhausting your L&D team, Boon covered it in how companies scale leadership growth without burning out their teams.
And measure the human outcome, not just the license count. Are people using the tool on real work? Has the quality or speed of that work changed? Are managers reporting that their teams have shifted? Those are HR's metrics to own, and they're the ones that tell you whether adoption is real. For the broader question of how to track this kind of impact, measuring coaching ROI is a useful starting point.
Frequently Asked Questions
Who owns AI adoption, IT or HR?
Both, but they own different halves. IT owns tool selection, security, and integration. HR owns the people side: behavior change, manager enablement, and whether people actually use the tool differently. Most companies hand the whole thing to IT, which is why adoption stalls. The people part has no owner.
What's the difference between deploying AI and adopting AI, and why does that gap stall rollouts?
Deployment means the tool is live and accessible to everyone; adoption means people actually use it on real work, regularly, in ways that change the output. Deployment is IT's win to claim, adoption is HR's job to deliver. Most rollouts fail because the company treats the two as the same: the tool goes live, licenses get assigned, and everyone assumes the job is done. But the real blockers, fear, trust, habit, and lack of manager buy-in, are human problems that no software rollout addresses, so usage flatlines after a few weeks.
How does coaching help with AI adoption?
Coaching works on the actual blocker instead of re-teaching features. A coach helps a person work through what's stopping them from using the tool, which is usually some version of "what does this mean for my job." That's a conversation, not a curriculum. There's more on this in Boon's piece on AI transformation coaching.
The Cost of Leaving Adoption Unowned
Leave the human half ownerless and the cost isn't a failed pilot. It's a slow, quiet stall: six weeks in, the dashboard says everyone has access, usage is flat, the CEO is asking why the AI investment isn't showing up in the work, and nobody can point to who owned the part that didn't happen. The ownership question was never actually answered.
HR can answer it. The work, changing how people work, is the work HR already knows how to do. Boon pairs the tool rollout with coaching at the manager and individual level, so behavior shifts instead of reverting, and managers get the language to model the change on Monday morning rather than after the next webinar. If your AI rollout is live and the usage isn't following, come talk to us about how it works. The tool was the easy part. The people part is the part you can still win.