Helping CTOs & CIOs turn AI ambition into an operating model: feedback loops, governance, and execution across people, process, technology | CAIO @ Zühlke
Joined December 2011
- Tweets 3,926
- Following 687
- Followers 794
- Likes 3,759
1,732 Photos and videos
1/
On Friday, one government letter switched off frontier AI for hundreds of millions of people overnight. No outage. No bug. A letter.
The US ordered Anthropic to block Fable 5 and Mythos 5 for every foreign national. Worldwide.
1
31
4/
Sovereignty is not autarky. It's not building everything yourself.
It's not being fully dependent on whether a foreign authority cuts your access tomorrow.
Loosely coupled. Multi-sourced. Exit paths by design.
1
7
5/
Before the next launch dazzles us, ask the boring question: if this gets switched off tomorrow, can your business still act?
Capability to act beats access to the frontier.
bit.ly/4g1R5vR
7
Jun 13
1/
Most teams are using AI to speed up the part that was never the problem.
Writing code was rarely the bottleneck. The delay sits before development: fuzzy goals, slow decisions, handoffs between silos.
1
18
Jun 13
4/
Teams are getting smaller too. The Agile team of seven plus or minus two collapses to three plus or minus two, each person backed by four or more agents.
But small only works if every person can think across the whole product, not just their own slice.
1
6
Jun 13
5/
The scarce skill is no longer writing code. It is judgment about what to build and why.
Full conversation on the Hello 50:50 World Podcast:
buff.ly/5RwhMJV
2
Jun 12
28
Jun 11
1/
The biggest risk in enterprise AI is not that it fails.
It is that it works, and then the bill arrives.
Just did a podcast on the part of AI nobody puts on a slide: token economics.
1
17
Jun 11
4/
Two things protect you.
Stop counting tools. That is procurement, not adoption.
And fix the broken process first, or AI just multiplies your mistakes faster.
1
1
Jun 11
5/
Past the pilot stage and trying to make AI pay off at scale? This episode is for you.
🎧 buff.ly/0zvhkhn
2
Jun 10
1/
The hottest job in AI is just a DevOps engineer who learned RAG.
Someone read 889 "AI Engineer" job ads from January 2026. Only 4.4% are research. 95.6% are production.
The data is hard to argue with.
1
37
Jun 10
4/
This is why "we just need better prompts" was always wrong.
The market does not pay for clever prompts in a vacuum. It pays for system design you can measure and trust in production.
The strongest industries? Finance, healthcare, cybersecurity, legal.
1
6
Jun 10
5/
The next competitive edge is not AI itself. It is the engineering around it.
So are you hiring for prompt tricks, or for the discipline to run AI in production?
Study (889 job descriptions): buff.ly/XOExLqr
6