The conversation HR and leadership need before AI training begins.

I get the call almost every week. A leadership team or HR director reaches out because they want AI training for their company. They have heard the team needs to be more efficient. They have read about productivity gains. They want to bring someone in. They want it scheduled in the next month.

I always ask the same question first: what conversation has leadership and HR had about what AI means for the workforce? Almost every time, the answer is some version of "we haven't really gotten into that yet." The training is being scheduled before the strategic conversation has happened.

That's the order most companies use, and it's exactly why so many AI training initiatives fall flat.

Why the order matters

When training comes before the strategic conversation, employees fill in the blanks themselves. And what they fill in is usually fear.

They sit down for the workshop and the unspoken question hanging in the room is: am I being trained to be replaced? No matter how skilled the trainer, that question shapes how every demonstration lands. People take notes politely and then go back to their desks and don't open the tools again. The training was theatre. The real conversation, about whether their job is safe, was never had.

Companies that get this right do the opposite. Leadership and HR sit down before any training is scheduled and align on what AI actually means for the organization. Then they communicate that to the team. Then they bring in training. The training works because employees walk in already trusting the company has thought about them.

The conversation itself

This conversation isn't a one-meeting checkbox. It's a series of decisions leadership and HR have to make together, then communicate clearly. Here are the questions that need real answers before training begins.

1. What is AI for, in our specific business?

"Efficiency" is not an answer. It's a word that means different things to different people, and that vagueness is exactly what makes employees nervous. Be specific. Are you using AI to handle the volume of work you can't currently get to? To free up senior people for higher-leverage work? To reduce the time it takes to ship things? To eliminate errors in repetitive tasks?

The specifics matter because they tell employees what kind of value AI is meant to create — and by extension, what their role becomes in that future.

2. Which roles are we expecting to evolve, and how?

Don't dodge this. If AI changes how a role works, leadership needs to think through that change before the training, not after. The marketing manager whose job becomes 60% strategy and 40% AI orchestration is doing different work than the marketing manager who used to spend most of her time producing assets. That shift needs to be acknowledged, planned, and supported.

Some roles will change a lot. Some will change a little. Some genuinely won't change. Get clear on which is which before you train.

3. Are any roles at real risk of being eliminated?

This is the hardest question and the most important one. If the honest answer is yes, you have to plan for that. Not by pretending it isn't happening, but by deciding how you'll handle it: redeployment into other roles, severance support, retraining pathways, or transparent communication about timing.

Companies that try to hide this from employees never actually hide it. People feel it. The trust erosion is much worse than honest, well-handled change.

4. How do we want employees to feel about this?

The answer leadership wants to give is "excited." The honest answer is usually "informed, supported, and clear about what's happening." Those are very different goals, and the second one is what builds adoption.

Employees don't need to be enthusiastic about AI. They need to feel that leadership has thought about them, communicated honestly, and given them the tools and time to adapt. That foundation is what makes training stick.

5. Who owns institutional knowledge, and how do we protect it?

This is the question almost no one asks at this stage. When AI changes how work gets done, the senior people who carry years of unwritten organizational knowledge become both more valuable and more at risk. More valuable because their judgment is what makes the AI useful. More at risk because if AI handles execution, leadership starts wondering if they can run leaner at the senior level.

Don't make that mistake. Identify the people whose departure would create real gaps and make sure your AI strategy treats their knowledge as something to capture and amplify, not something to replace.

"Employees don't need to be enthusiastic about AI. They need to feel leadership has thought about them, communicated honestly, and given them tools and time to adapt."

How to communicate the answers

Once leadership and HR have aligned on the questions above, those answers need to be communicated to employees before training is scheduled. Not in a vague all-hands. In a clear, structured way that gives people what they need to engage.

1. Start with leadership's framing. Why is the company adopting AI now, in plain language. What problem is it solving, and what isn't it solving.
2. Be honest about what's changing. Which functions, which workflows, which expectations. People can handle change. They struggle with surprise.
3. Acknowledge the question on everyone's mind. Job security. Even if the answer is "we don't expect role eliminations from this rollout," say it. Don't make people guess.
4. Explain the support structure. Training, time to learn, office hours, internal champions, and what employees should do if they're struggling.
5. Then schedule the training. Now the workshop has the right context to land.

What to do if you've already done it backwards

If you're reading this and realizing you scheduled training before the conversation, don't panic. You can still recover the situation, but you have to do the conversation now, retroactively.

Pause any further training rollouts. Have leadership and HR meet to align on the questions above. Communicate the answers to the team transparently, including a brief acknowledgment that the order was off. Then re-engage with training, ideally with the trainer adjusting the approach to reflect the new context.

Employees will respect the correction more than the original mistake. Trust gets rebuilt by acknowledging missteps and making them right, not by pretending they didn't happen.

The real point

AI is not a software rollout. It's an organizational change. Treat it like a software rollout — schedule the training, hand out the licenses, hope for the best — and you'll get the surface-level adoption that fades by month three.

Treat it like the organizational change it is — start with leadership and HR alignment, communicate clearly, sequence training around a real workforce plan — and you'll get the deep adoption that actually delivers ROI.

The conversation has to come first. Everything else hinges on it.