What AI Actually Does to a Team

The conversation about AI in organizations almost always focuses on the wrong thing.

Executives want to know about productivity gains. How much faster will my team work? How many tasks can I automate? What's the ROI? These are reasonable questions, and they have real answers. But they're not the most interesting thing that happens when you introduce AI to a team.

What's more interesting (and what I've watched happen across enough organizations to recognize the pattern) is what it does to the people. To the dynamics. To who becomes valuable and who feels threatened and what kinds of conversations suddenly become possible.

The fear arrives before the tool

In every training I've done with teams (not executives, teams) the first thirty minutes are always the same. People are polite. Attentive. Taking notes. Asking good questions.

And also, underneath all of that: afraid.

Not afraid of the training. Afraid that the training is a prelude to something else. That the reason their company is investing in AI education is to figure out who they no longer need. That the efficiency gains will mean headcount reductions. That "learning AI" is actually code for "making yourself redundant."

This fear is almost never acknowledged out loud in the session. But you see it. In the questions people ask, in how they frame their use cases, in the way certain participants go very quiet when the conversation turns to task automation.

The first thing I do, before any tool, any technique, any framework, is address it directly. I say it plainly: AI doesn't replace the people who understand your business. It replaces the task, not the role. And the people who understand how to direct AI will become more valuable, not less.

I mean it. And I can show them why I mean it, with examples. The evidence helps. But the willingness to acknowledge the fear first is what makes the rest of the session land.

Who surprises you

Every team has someone who turns out to be exceptional at working with AI. And it's almost never the person you'd predict.

The person who is spectacular with AI tools is not always the most technically fluent. It's not always the youngest person in the room. It's not always the one who was most enthusiastic about the training.

It's usually the person who is the most specific thinker. Who communicates precisely. Who, when they don't like an output, can articulate exactly what's wrong and what they want instead. Who has the patience to iterate without getting frustrated by imperfect first results.

These are qualities that have value in all kinds of work, but AI makes them uniquely visible. In a normal workflow, the person who writes crisp briefs and gives clear feedback is valuable but not dramatically differentiated. When working with AI, they're an order of magnitude more effective than someone who can't articulate what they want.

Every training I run, I watch for who this person is. They don't always know it yet themselves.

What changes that nobody expected

Here's what surprises teams most about working with AI: it improves how they communicate with each other.

To use AI effectively, you have to be precise. You have to articulate what you want, what constraints matter, what good looks like. If you've been vague in your briefings (with the AI, with your colleagues, with your clients) the AI will make you confront that vagueness immediately. A vague prompt produces a generic result. You can't hide behind "you know what I mean."

I've watched teams come out of AI training and give each other better briefs. More specific. More contextual. More willing to define what success looks like before starting the work. The AI forced the discipline, and the discipline migrated to the human interactions.

This wasn't the training objective. It's a side effect. But it might be more valuable than the productivity gains.

The knowledge distribution problem

One of the most consistent applications I see landing well across organizations (regardless of sector) is knowledge capture.

Every organization has experts. People who carry critical knowledge in their heads: processes, exceptions, institutional history, client context. This knowledge is fragile. It lives in one person, and when that person leaves, takes vacation, or simply isn't available, the knowledge isn't either.

AI doesn't create the knowledge. But it creates a structure that makes capturing and distributing it possible at a scale that was previously impractical. When you can ask a language model a question and get an answer drawn from your own documented knowledge base, the distribution problem changes. The expert's knowledge becomes accessible without requiring the expert.

This is where I see the most genuine relief in teams. Not excitement: relief. The recognition that the thing that has always been fragile doesn't have to be.

After the training

The teams I worry about are not the ones who leave the session skeptical. Skepticism is engaged. It means they're thinking about it.

The ones I worry about are the ones who leave enthusiastic but unsupported. Who go back to an organization where leadership has bought into the idea of AI but hasn't committed to the follow-through: the time to experiment, the permission to integrate new tools into workflows, the acceptance that the first month will be slower before it gets faster.

AI adoption in a team is not a training problem. The training is necessary but not sufficient. What happens in the weeks after the training determines whether anything changes.

The organizations where it sticks are the ones where someone in leadership models the use: where a manager or director shows up to a meeting having used AI in their own preparation and talks about it openly. Where there's a space to share what's working and what isn't. Where the inevitable awkward first attempts are treated as information rather than failure.

That's not a tool problem. That's a culture problem. And it's the one I can't solve in a session.

But I can point at it. And I do.