Gradient blur
Blog
Artificial Intelligence
Learning Solutions
Back to overview

Where are your people with AI?

Measure first, then choose a training. Johan Peeters explains why a conversation tells you more than a score.
28 - 10 - 2026

Johan Peeters is the Business Unit Manager for Learning Services at Xylos. He helps organizations get their people on board with AI. In this blog, he explains how to assess your employees’ AI proficiency without having to survey each person individually.

You’ve rolled out Copilot licenses and scheduled training. Yet only some of your employees are using AI in their day-to-day work. You won’t know what training the others need until you understand where they stand today.

Which AI applications are relevant, and which are hype?

I think the honest answer is that you don’t know that in advance. What I do know is this: AI doesn’t replace learning. AI makes learning faster and even more important. Working with AI is becoming the norm today, so the real question is how to use it to work more efficiently, productively, and effectively.

In 2024, I helped build Xylos’ AI innovation lab, where we launched more than fifteen new services. My biggest lesson from that period? Technology only delivers value if people actually use it.

At the organizations we work with today, I see the same pattern time and again. Some colleagues use Copilot every day, while others have never even tried it. The question I’m asked most often is: what kind of training do they need? My answer is always the same: you only know once you’ve mapped their knowledge level.

Why do some colleagues use Copilot daily and others never?

The tool is the same for everyone. The difference lies in what each person is already capable of and recognizes. These reasons come up time and again when we assess users’ knowledge levels:

  • They don’t see how Copilot can be useful in their own job. An HR professional uses it differently than a project manager, and a general demo rarely makes that connection.

  • They tried it once and gave up. If you ask Copilot a vague question, you’ll get a general answer. Those who don’t know how to refine their query quickly decide it’s not worth their time.

  • They’re unsure whether they should or can trust Copilot. As long as there are no clear guidelines on what data is allowed, they’d rather stay away from it.

Why doesn’t one-size-fits-all AI training work?

A standard training course is too simple for those who already work with Copilot every day. For those who have never written a prompt, it moves too fast. And those who are unsure about the rules need different help than those who want to know how Copilot works with the rest of Microsoft 365. If you put all those people in the same session, the training will miss its mark for most of them.

If you don’t know your team’s skill level, you’ll choose a training course based on a hunch. Assessing their knowledge level replaces that hunch with an objective answer.

What do you need to know before choosing AI training?

I always look at four things:

  • The level: is someone already working with Copilot, or are they just getting started?

  • The tasks: how does someone start their workday, and which tasks come up every week?

  • Needs: where does someone get stuck, and where do they have doubts?

  • Learning style: do they prefer a classroom session, short videos, one-on-one guidance, or a game?

How do you measure your employees' AI skill level?

You have two options: a questionnaire or an interview. I’d go with the interview.

A questionnaire gives you a score. It indicates how far along someone is, but says little about why someone needs AI in their own work.

At Learning Services, we have an agent working for us. Her name is Eva. Eva talks to our clients’ employees via a Teams call, using plain language and at a time that fits their schedule. An AI agent that talks to your people? Yes, really. Based on those conversations, she assesses each employee’s level of knowledge, along with their tasks, use cases, and needs.

We’ve been down this path ourselves. We started with questionnaires, switched to a chat interface, and now work with Eva, who talks to people the way a person would. This gives you an overview based on a real survey.

I explain how Eva works in this video.

The conversation has another benefit. Everyone who takes part gets their first hands-on experience with AI. For employees who’ve never worked with AI before, that’s already a first step.

What does the AI agent ask your employees?

These are open-ended questions about your employees’ work and responsibilities. The conversation is brief. For example:

  • What does your typical workday look like?

  • What’s the first task you tackle in the morning? Reading emails, preparing for a meeting, …

  • How do you interact with customers, suppliers, and partners?

  • How do you process the information from those conversations, and how does it get shared with your colleagues?

From those answers, you’ll gain more than just a score. You’ll see which AI applications are already in use on the job, where work can be done more efficiently, and how someone feels about AI.

What do you do with the results?

We match the report with the available learning solutions. This gives you a personalized learning path for your organization. We’ll discuss this with HR or the learning department: what are you addressing right now, and where do you want to invest further?

If a team needs basic digital skills, that’s where you start. If other colleagues are ready for real-world applications from their own work, they can build on that. This creates a growth path tailored to each group, rather than a training calendar that’s the same for everyone.

Assess progress again after the program. This will show you whether skill levels have improved and where adjustments are needed. To learn how to design a program that sticks, read our blog on training that pays off. Don’t have your own L&D department? This article explains how to handle that.

The case study of Maxeda DIY Group (the parent company of Brico and Praxis) shows what such a program looks like in practice. Their AI literacy program grew to include more than seventy workshops, and today the team uses Copilot every day.

My advice? Start by measuring. Only once you know where your people stand can you choose the training that will help them move forward.

How Xylos Assesses Employees' AI Proficiency

Xylos is a Belgian family-owned tech company headquartered in Antwerp. Training has been our specialty for forty years.

With Xylos Insights, our AI agent Eva conducts a brief conversation with each employee. You’ll receive a score for each employee, a heat map for each team, and an overview of the AI applications already in use on the job. Use it as a standalone quick scan, or as the starting point for a learning journey such as the AI literacy program.

Ready for a conversation?

Curious to know where your employees stand on AI? Schedule a no-obligation consultation with Johan and his team.