Governance and responsible adoption of AI in high-stakes industries. Est. 2026
AI researcher and builder

I work on how AI gets adopted and governed in industries where the stakes are real.

Mostly healthcare, public services, and law. I write here about what I am figuring out as I go, usually what holds up in practice and where it is easy to get wrong.

Pedram Madani
2026 · 09 · 04

Shipping is where the work starts

Go-live is treated as the finish line. It is the starting line. Most pilots do not die at launch, they die quietly in the months after, because the day they went live they became nobody's job.

Published
2026 · 08 · 29

The EU AI Act, in plain terms

The regulation is long, and most of the fear around it is misplaced. For almost any organisation the practical question is small: which tier are you in, and what does that actually require of you?

Published
2026 · 07 · 11

Most AI advice can't be wrong

A lot of what you are told about AI is built so that nothing could ever prove it false. That is not insight. It is trend-following in a suit, and there is a simple test for it.

Published

Newsletter

By email

Read it here, or have it reach you. New essays and short notes by email, from inside real deployments. Infrequent, no filler, unsubscribe in one click.

Contact

By request

Working through a real AI decision? Tell me what you are deciding and I will reply if I think I can help. Not a sales call.

Let’s have a chat →