Almost everyone I talk to has sat through the same meeting.
There's an all hands, or a town hall, or a post from someone senior. The company has an AI strategy for workforce data now. There's a tool, or a pilot, or a partnership, and probably a slide with a productivity number on it. People nod. Some are excited. Some are quietly unsettled. Then the meeting ends and everyone goes back to their inbox.
Six months later, ask those same people whether AI has actually changed how they work, use data, or get to insights. Most of them say no. Not really. A few use it to clean up emails. Somebody on the analytics team is doing something interesting with it that nobody else has time to learn. The work looks about the same as it did before the announcement.
The announcement and the change are not the same event.
I've spent the better part of eighteen months paying attention to that gap. Partly because it’s my job. I work in People Analytics at One Model, which builds data modeling software for HR teams, so I get to watch both sides of enterprise AI adoption. I see what companies announce. I see what their people actually do on a Tuesday. The distance between those two things is wider than most of us say out loud.
That distance isn't a failure of nerve. There are real forces holding it open, and three of them are worth naming.
The first layer is the most legitimate and the least discussed. Before a large company can put an AI model anywhere near its data, somebody has to answer hard questions about where that data goes, who can access it what happens in a breach, and which regulator asks about it first. Those questions have real answers. Finding them takes months.
This gets characterized as bureaucratic drag. It isn't. It's the organization behaving correctly. But it does mean the person who watched the all hands in March may not have access to anything until October, and by October the AI tool they were promised has been replaced twice.
The second layer is the one most companies mistake for the whole thing. Getting a tool procured, secured, and provisioned is a project with a finish line. Getting people to change how they work has no finish line, and it doesn't respond to provisioning.
Most rollouts I've watched treat training as the bridge between approval and adoption. An hour long session, a recording, a prompt library on the intranet. Then adoption gets measured in logins, logins look fine for three weeks, and everyone declares it done.
If that’s where you’re looking, you’re measuring the wrong KPI; logins aren't adoption. Somebody opening a chat window once a week to reformat a paragraph is not a person whose work has changed.
The third one is uncomfortable, so it mostly goes unmentioned.
When you stand back and examine what employees bring to the table, much of their value rests on deep, human domain expertise–call it "tribal knowledge." You're the person who knows how the comp cycle actually runs, or which fields in the HRIS are trustworthy, or why the headcount number in that report is wrong. That knowledge is why people come to you. It's a real part of why you have the job.
AI is unusually good at flattening exactly that kind of advantage. Not perfectly, and not yet everywhere. But enough that if you've built standing on being the person who knows, there's a quiet incentive not to go first.
Nobody experiences this as self interest. It shows up as reasonable caution. "The output isn't accurate enough." "It doesn't understand our data." Sometimes that's true. Sometimes they’re the words of someone who thinks AI will dull their competitive edge
Put those three together and you get a pattern that repeats across industries and company sizes. A pilot launches. It goes reasonably well. Somebody writes up the results. And then the structural weight of the organization quietly reasserts itself, and six months later the work looks about the same as it did.
Here's the part I find genuinely useful.
If you've been feeling behind, you're probably not. The gap between announcement and change is measured in years, not quarters, and almost nobody is on the far side of it. The pace of what the technology can do is real and it is fast. The pace at which organizations absorb it is neither of those things. You live on the second line, along with everyone else you know.
That's more time than the noise suggests. It is not infinite.
What I've come around to is that the hard part was never the technology. Organizations have gotten good at redesigning workflows. The redesign they keep skipping is the people, and that one doesn't respond to a project plan.
I'm giving a webinar with CultureCon on August 27th about what that actually takes, for individuals first and for leaders second. There's one question I plan to ask the room live, and I'd rather you sat with it now than heard it from me cold.
Where are you, honestly, on your own AI adoption curve? Not where your company says it is. Where you are.
I'm curious what you find.
Register for "Get Your AI Footing: The Ground Has Shifted" — Thursday, August 27, 11:30 AM CT
Separate Version for Culture Con
I’ve sat through the exact same meeting as everyone else.
There's an all-hands or a memo from leadership. The company has an AI strategy for HR data now. There's a new pilot or a software deal, and usually a slide with productivity promises. People nod. A few are excited. Others are quietly anxious. Then the call ends, and everyone goes back to their inbox.
Six months later, ask those same people if AI actually changed how they work or analyze data. Most say no. Not really. Someone uses it to polish emails. A person in analytics is trying something cool that no one else has time to learn. Otherwise, the work looks identical to last year.
An announcement and real change aren't the same event.
I've spent eighteen months observing that gap.
Partly because it's my job. I work in People Analytics at One Model, building HR data tools. That gives me a front-row seat to both sides of this. I see what executives promise in press releases, and I see what people actually do on a Tuesday. The gulf between those two things is substantial.
That gap isn't a failure of effort.
Real structural forces hold it open, and three of them stand out.
The first factor is the most legitimate. Before any large company plugs AI into workforce data, someone has to answer tough questions. Where does the data go? Who can see it? What happens in a breach? Finding those answers takes months.
People call this bureaucratic drag. It isn't. It's the company behaving responsibly. But it means someone who watched a launch presentation in March won't get access until October. By then, the promised tool was replaced twice.
Buying and securing software has a finish line. Getting people to change their daily habits doesn't.
Most rollouts treat training as the bridge. You get a one-hour webinar, a recording, and a prompt list on the intranet. Then leaders track logins for three weeks, see a spike, and call it a success.
Logins aren't adoption.
Opening a chat box once a week to rephrase a sentence isn't changing how you work.
People don't like to talk about this one.
Much of an employee's value comes from deep, personal domain knowledge. You're the person who knows how the comp cycle actually works, or which HRIS fields are broken, or why a specific headcount report is wrong. That knowledge is why people rely on you. It's why you have a job.
AI is good at flattening that exact advantage.
If your standing relies on being the person who knows, you have a quiet incentive not to go first.
People don't frame this as self-preservation. It shows up as reasonable caution. They say the output isn't accurate, or that it doesn't get their data. Sometimes that's true. Sometimes it's just someone protecting their competitive edge.
Put those three together and the pattern repeats everywhere.
A pilot launches and goes fine. Someone writes up the results. Then the weight of the company takes back over, and six months later nothing changed.
Here's the part I focus on.
If you feel behind, you aren't.
The distance between an announcement and real change takes years to cross, not quarters. Almost nobody is on the other side yet. The technology moves fast, but organizations absorb it slowly. You live on the second line, along with everyone else.
That gives you time, even if it isn't infinite.
The tech was never the hard part. Companies know how to redesign processes. They keep skipping the people side, and that part doesn't react to a project plan.
I'm doing a webinar with CultureCon on August 27th about what this actually takes.
There's one question I'm going to ask the room live, and you should probably sit with it now.
Where are you, honestly, on your own AI adoption curve? Not where your company says it is. Where you are.
I'm curious what you find.
Register for "Get Your AI Footing: The Ground Has Shifted" (Thursday, August 27, 11:30 AM CT).