Johan Elias was interviewed by Kanaal Z / Trends Z at Educam in Evere on 9 June 2026, following the publication of the FPS Economy study on AI adoption among Belgian companies.
On Friday I received a call from a Kanaal Z journalist asking whether I was available on Monday for an interview. The reason: a new study from the Belgian Federal Public Service Economy on AI adoption among Belgian companies. We agreed to meet at Educam in Evere, one of our clients, where I had just scheduled an AI Champions session. That way, we could connect theory directly to practice, live.
That combination — a study measuring the state of AI in Belgium and a client actively working with it — set me thinking. The figures are encouraging. And yet they tell only part of the story.
The FPS Economy published new figures on AI adoption among Belgian companies. In 2024, 24.71% of Belgian enterprises used AI; by 2025 that figure had risen to 34.54%. Belgium ranks in the European top five, above the EU27 average of 19.95%. Good news, then.
And yet.
Anyone who reads the study more carefully encounters a different reality. The most commonly used applications are text analysis, language generation and image generation. The most popular domains are business administration, accounting and marketing. These are not poor applications, but they reveal where most companies stand today: in wave one.
Wave one: AI literacy and low-hanging fruit
Over the past three years, Rescope has guided dozens of mid-sized and large organisations through their first steps with AI. What we consistently observe aligns closely with what the FPS Economy study measures. Companies start with out-of-the-box tools: Microsoft Copilot, ChatGPT, Gemini. They configure, experiment, they save time on email, reports, translations.
That first wave has real value. Employees build AI literacy. Workflows become sharper. Teams discover what is possible. And the AI Champions, the internal bridge-builders between IT and business whom Rescope helps to develop, ensure that knowledge does not stay with one enthusiast but spreads throughout the organisation.
But it is wave one. The beginning of the story, not the middle.
The picture the study shows is yesterday's
What the FPS Economy study measures is a snapshot of what companies did in 2024 and 2025. That is valuable. But digital workers — AI agents that independently execute complete processes without a human approving every step — already exist. Not as prototypes, not as promises. They are operational, at companies that chose not to wait.
At the start of this year, Rescope clients began working with agentic AI: agents that process incoming invoices, compile reports from multiple sources, prepare dossiers for account managers. While the market as a whole is busy with 'AI for texts', a group of companies is already engaged with wave two.
The gap the study reveals between large companies (76.41% AI use) and small companies (28.82%) is real, but less alarming than the gap absent from the study: the one between companies using AI as a better search function, and companies using AI to fundamentally redesign their processes.
Wave two: from automating to transforming
Agentic AI is not simply faster automation. It is a different way of organising how work happens. Processes are not copied and accelerated — they are redesigned from the ground up. Roles change: employees orchestrate, adjust, make decisions based on what agents deliver. The administrative layer largely disappears. What remains is judgement.
That asks something of leaders that goes beyond purchasing a licence. It requires an understanding of where the boundaries lie, what is responsible, which processes are suitable and which are not. Those who are not actively thinking about this today risk optimising existing processes while their business model is already under pressure.
The primary barrier identified in the study — lack of expertise (74.53%) — is precisely the point. Not technical expertise, but strategic expertise: the capacity to see which bets are worth making.
Organising ownership is not something you can outsource
Wave two is not something you can delegate to a technology vendor. The digital complexity that agentic AI brings with it requires organisations to understand themselves what they are building and why. Champions who can guide agents. Leaders who make the right strategic choices. Internal academies that anchor knowledge so it does not leave with one employee.
That is precisely why Rescope is launching a new platform this autumn connecting all stakeholders: users, AI Champions, steering committees, leaders, AI experts and data engineers. Not because technology is the answer to everything, but because collaboration around technology is the only way to structurally anchor ownership.
The real lesson
The FPS Economy study shows that Belgium is on track for wave one. That is good. But the European target of 75% AI use by 2030 does not measure what it should measure. Adoption is not an end point. The question is not whether your employees use AI, but whether your organisation changes through how they use it.
Wave two is not something to consider later. It is already under way. The organisations that are deliberately thinking about it now — which processes are suitable, how to bring their people along, how to anchor ownership — are the ones that will not be playing catch-up three years from now.
Source: FPS Economy, study on AI use at Belgian companies (2025). Johan Elias was interviewed by Kanaal Z / Trends Z on 9 June 2026 following the publication of this study.