Judgement does not automate.

I started by writing code. Front-end, more than ten years ago, when the job was turning someone else's decision into a screen that worked. It did not take long to realise the screen was the least of it. What mattered was the decision behind it: what the user was asked for, what was hidden from them, what was assumed about them. That is where I stopped being someone who executed decisions and became someone who questioned them.

That is the path that brought me here. From building to designing, from designing to leading product, and from leading product to working in the one place where something real is now at stake: the boundary between what a person decides and what a machine decides.

I have designed and led product in healthcare, finance, real estate and energy. Sectors where the margin for designing badly is tiny, because an automated recommendation can affect a course of treatment, someone's savings or a city's power supply. Working that close to real consequences changed what I pay attention to. Whether a model is impressive matters less to me than what happens when it gets it wrong, who answers for it, and whether the system leaves the person better or worse off than before.

My thesis is uncomfortable for much of the industry: most of the AI problems I see in organisations are judgement problems. Teams optimise the model, the prompt, the latency, and defer the hard question of whether that system should be making that decision at all. Neither engineering nor compliance can answer that question for you. It is about what kind of product, and what kind of organisation, you want to be.

To answer it I wrote PRAXIS. A framework for deciding what is delegated to the machine and what stays human, built on a tradition that takes human agency seriously: Arendt, on what happens when we stop thinking because the system thinks for us; Jonas, on a responsibility proportional to our power to act; Murdoch, on attention as a moral act in itself. I cite them because product design has spent decades making ethical decisions without naming them, and AI has made that impossible to keep getting away with.

Today I work where three things meet that most people treat separately: product strategy, system behaviour and applied ethics. I treat them as one discipline. A product that decides for you has a character, whether its team intended one or not. And character is designed.

I also lead teams. I have run product and design organisations of up to fifteen people, set roadmaps with business, operations and legal at the same table, and launched new business lines alongside a live platform. If you want help deciding, with judgement, where the machine adds and where it subtracts, let's talk.