Insights

What we notice in the practice of AI at mid-sized organizations.

4 September 2026 · Michael Blok

Three AI outages, one lesson about vendor risk

ChatGPT, Claude and Grok went down at the same time. That does not make AI riskier overnight, but it moves vendor risk from theoretical to observed.

31 August 2026 · Hans-Peter Harmsen

Everyone has AI tools, nobody has an AI strategy

Twelve departments with twelve AI tools is not a strategy. What is missing is not the thirteenth tool, but one shared memory that all of them draw from.

3 August 2026 · Hans-Peter Harmsen

How much do you leave to AI agents? Two questions

Trusting agents more because the models are getting smarter is like unbuckling your seatbelt in a nicer car. Two questions decide it, and both are about the task.

30 July 2026 · Michael Blok

Five AI risks SMEs actually need to manage

Vendor, economic, legal, strategic and implementation risk. Most AI risk is ordinary business risk in a robot suit, and an SME has no compliance department.

25 July 2026 · Michael Blok

Would an AI crash actually hit your business?

AI lab valuations are unsustainable, so something will give. But a crash barely touches an SME with working use cases. Pricing risk is the thing to watch.

21 July 2026 · Hans-Peter Harmsen

The AI bill, and the router that shrinks it

AI at scale gets expensive, especially with agents. A model router picks the cheapest model that will do per call. What it saves, and where the catches are.

18 July 2026 · Michael Blok

What should my daughter study in the age of AI?

Four criteria for choosing a degree while AI eats the map: be interesting, judge well, own things, be accountable. And where that leaves medicine and coding.