AI adoption is a people problem, not a technology problem

Most companies don't have an AI problem. They have a Tuesday morning problem. The new tool is paid for, the launch email went out, and people still open the same old spreadsheet.

That's not laziness. That's being human. And it's why we believe AI adoption is a people problem first, and a technology problem a distant second.

Where this belief comes from

AK, our founder, led AI adoption work at Google as Executive Sales Coach Lead during the ChatGPT and Gemini launch era. That's a front row seat to one of the biggest shifts in how work gets done, inside a company that builds the technology.

The lesson he carried out of that seat is simple. Tools roll out. Behavior doesn't change on its own. And when behavior doesn't change, the investment stalls.

Not because the tool is weak. Because nobody planned for the humans who were supposed to use it.

The research points the same way

You don't have to take our word for it.

In its AI at Work 2025 study of more than 10,600 workers across 11 countries, Boston Consulting Group found that only 51% of frontline employees are regular AI users, and that number has stagnated. Only 36% of employees said they feel adequately trained. Just 25% of frontline workers said their leaders give them enough guidance on AI.

BCG also puts the value math plainly. By its estimate, only 10% of the value in an AI transformation comes from the AI application itself, and another 20% comes from the underlying data and technology. The remaining 70% comes from things like workflow redesign, culture, governance and how people and AI work together.

Sit with that for a second. Most of the value has nothing to do with which tool you picked. It's about people and how work actually gets done.

Why adoption stalls

In our experience, the same four things keep showing up.

Habits. People have a groove. The old way feels fast because they've done it a thousand times. Any athlete knows this one. Change your running form, and the moment you get tired, your body falls right back into the old stride. Work is the same. Under deadline pressure, people reach for what they know.

Fear. A new tool can land like a quiet message: we're trying to need you less. In that same BCG study, 41% of respondents worried their roles could disappear within the next decade. You can't train your way past a fear nobody has talked about.

Unclear workflows. "Use AI more" is not a plan. Which task? Which step? What does done look like? If nobody can point to the exact moment in someone's day where the tool fits, it won't fit.

No one owning the change. The software gets an owner. The behavior change usually doesn't. The tech team ships it, leadership announces it, and everyone assumes someone else is coaching people through it. Usually, nobody is.

What actually works

Start with people. Before anyone opens a tool, ask your team what eats their week. What they dread. What worries them about all this. People support what they help shape, and their answers tell you where to begin.

Map the workflow. Write down how one real job actually gets done, step by step. Not the org chart version. The real one, workarounds and all. Then find one step where help would make a real difference.

Build small reps of new behavior. Nobody fixes a race in one practice. AK was a four time All American at the University of Oregon, and none of that came from one big workout. It came from reps. Small, repeatable, a little better each time. Pick one task, practice it together, and celebrate the first win out loud. BCG's study points the same way: people who got five or more hours of training, especially in person and with coaching, were significantly more likely to become regular users.

Measure what people actually do. A survey that says people "like the tool" tells you very little. Look at behavior. Did they use it on the real task this week? Did they come back? Where did they stop? What people do tells you the truth faster than what they say.

That last one is close to our hearts. Paying attention to what people do, not only what they say, is how you notice someone is stuck before they quietly give up.

People first, then process, platforms last

This is the order we work in, every time.

People first. We start with a deep dive into the business and the humans in it. Who does what, what they need, and what a good outcome looks like for each role.

Then process. We map the workflows, roles and outcomes, set clear objectives, and research what good looks like. We design around how people actually behave, not how a slide deck hopes they will.

Platforms last. Only then do we build. We launch with your people, not at them, and then we grow together.

Most stalled rollouts run that order backwards. Buy the platform, bolt on a process, and hope people follow. It's like buying the best spikes on the market and skipping practice.

How it feels when you get it right

For the person on the team, it's the difference between being handed a tool with a speech about efficiency, and being asked how their job really works and getting a small win in week one. One feels like a threat. The other feels like respect.

For the owner, it feels like relief. The thing you paid for gets used. The team feels a little lighter instead of a little more watched. And you stop wondering whether you bought the wrong software.

You probably didn't. The gap is usually the part with a heartbeat. That's good news, because people can change. They just need someone to start with them.

If you're rolling out AI and it isn't sticking, we'd love to hear what you're seeing and talk through where to start.