AI adoption in Flanders doubled. What the numbers really tell us.

Three measurements, three snapshots. The most recent AI Barometer, which mapped the situation in 2025 and was published in early 2026, marks a turning point, but also shows where companies get stuck, and why the next twelve months are decisive.

What looked like a fringe story in 2022 is mainstream today: nearly six in ten Flemish companies use at least one AI technology. A near-doubling in two years. It is an appealing figure, and it is accurate.

But behind that percentage lies a more complex story. Not about how many companies use AI, but about how it got in, and whether that path is sustainable. That story determines whether your organisation turns AI into leverage in the coming years, or misses the boat.

The evolution in three numbers

Since 2022, the Flemish Department WEWIS has conducted the AI Barometer, a large-scale survey of Flemish companies (2025 sample: 2,915 usable responses from 9,610 companies surveyed). The trend is striking:

• 2022: 26% of Flemish companies use at least one AI technology.
• 2023: 32% — a modest rise.
• 2025: 58.8% — a near-doubling in two years.

The engine behind that leap is generative AI. Text analysis rose from 14% to 41%. Language and code generation went from 8% to 38%. Image, video and audio generation (a new category in 2025) immediately stands at 33%. Nearly half of all Flemish companies (48%) now use generative AI.

The low-threshold nature of tools like ChatGPT, Claude and Gemini is clearly the accelerator. But low-threshold is not the same as strategic. Using AI does not yet mean AI delivers value.

What the numbers hide

The adoption rate says something, but not everything. The 2025 Barometer exposes three uncomfortable truths.

1. Employees take the initiative, not management. In half of all companies (49%), AI use comes mainly from individual employees. Only 28% is actively driven by management. In other words, AI is rarely the result of a strategic choice. It crept in from the bottom up, tool by tool, task by task.

2. One in five is shadow AI. 20% of adopters use AI without the knowledge or approval of IT or management. This comes with real risks: sensitive data leaving the controlled environment, security gaps, potential GDPR breaches and exposure under the AI Act, which is now in force, yet familiar to barely 43% of adopters.

3. Only 17% has truly scaled AI. Almost half of adopters are still in the experimentation phase. A third has integrated AI into a few processes. But only 17% has rolled AI out across a broader set of domains within the organisation.

The pattern underneath is always the same. AI was not decided, it got in. And the biggest barrier companies report themselves confirms this: it is not budget and not technology, but a lack of knowledge, skills and experience (72.5%), followed by the difficulty of assessing where AI is meaningfully applicable (63.7%). The brake on responsible AI is not unwillingness. It is uncertainty.

The good news, for those who go beyond experimenting

For companies that take the step toward integration and scaling, the impact is tangible. Two-thirds of adopters (65%) saw concrete business impact over the past year:

• 53% increased the quality of their business processes (+15 percentage points vs. 2023).
• 40% managed to reduce costs (+13 pp).
• 31% launched new or significantly improved products or services (+3 pp).

Moreover, more than half of adopters (52%) report higher labour productivity among highly educated employees. Those who turn experimentation into implementation, reap the rewards.

“In practice, there is a great deal of cold feet and a wait-and-see attitude when it comes to really pushing through and going beyond the occasional prompt to rewrite an email.”

— Geert Vromman

So between “we use AI” and “AI delivers structural value” lies a gap today. And that gap determines who accelerates with AI in the coming years and who falls behind.

Governance is not a brake. It is the condition for moving faster.

This is where the two stories come together. Scaling without direction increases the risks. Direction without scaling delivers no value. The organisations that will accelerate in the coming years are not the ones that experiment hardest, but the ones that let experimentation and governance grow together.

Governance is often seen as a brake on innovation. In practice, the opposite is true. Clear agreements remove uncertainty. When employees know which tools are approved, which tasks are appropriate and where the boundaries lie, they can work faster without exposing the organisation unnecessarily.

That is precisely why AI today is first and foremost a governance question, and only then a technology question. The technology is here. It is accessible, powerful and cheap. What is missing is the structure around it: clarity on roles, visibility into where AI influences processes and decisions, and a shared framework that turns loose tools into a managed capability.

Do you know where your organisation really stands?

The question for the coming twelve months is no longer whether you use AI. Your organisation almost certainly already does, possibly more than you think, and in places you do not yet have in view. The question is whether that use is strategically anchored, safely framed and aimed at value.

With the CROPLAND AI Maturity Scan, we bring structure to that complexity. We analyse where your organisation stands across the four dimensions of our framework: Discover, Align, Design and Deliver, including visibility into shadow AI, data risks and your position under the AI Act. Not a one-off snapshot, but a compass for your next phase.

The technology is in. The question is whether you decide how it moves forward, or whether others decide that for you.

Ready to have that conversation? We are your navigator.

Source: Standaert, T., Lecocq, C., Andries, P. & Evens, T. (2026). AI Barometer: Adoption and use of Artificial Intelligence in Flemish companies — situation 2025. Department of Work, Economy, Science, Innovation and Social Economy (WEWIS) / ECOOM, Flemish Government. Earlier measurements: situation 2022 (publ. March 2023) and situation 2023 (publ. April 2024).