Positions worth defending

In care software, an overwhelming or confusing UX costs accuracy in a record someone relies on. These are areas missed by compliance checks or technical audits, but understanding and reflecting user mental models is a critical aspect of design in care.

Low-light rooms. Low-signal environments. Low tech proficiency across the people using this every day. None of these are a nice-to-have - it's how you stop errors before they reach a hazard log.

Treat accessibility as an inclusion checkbox and you design for the average case. Treat it as clinical safety and you design for the conditions people are actually in when the stakes are highest.

Accessibility, error prevention, how risk is surfaced etc. - these are all challenges which should be solved once and inherited, not decided differently in every flow. Don’t make your users re-learn an unpredictable rhythm.

The interesting question about AI in design is not "can it build the thing", but instead "what does it need in order to be trustworthy enough to build on." At Nourish, AI-assisted builds hit 70–80% first-pass quality using our design system - not because the model got smarter, but because Pulse became mature enough to give AI a reliable foundation to work from.

Skip that groundwork and you get fast output you can't trust. Would you rather spend a quarter strengthening a design system or a quarter cleaning up what AI built from a shaky one?

Permissions, provenance, consent, escalation, human override: these repeat across a product portfolio. Treat each as a fresh problem and you get inconsistent, quietly worse decisions. Design it once and every feature inherits the same discipline.

New regulation changes what "good" looks like in the interface: what gets documented, escalated, or made reversible. Reading that early and translating it into design decisions is a leadership skill, not something to wait on compliance to hand down.

That polished AI-generated prototype at kickoff looks like progress but skips the discovery that should come first, much less the organic cog-turning which occurs in a designers’ mind as they’d move within the canvas. We’re at risk of losing the brilliance of divergent thinking, of true ideation, of sketching several directions before they converge. This is often where my best work has emerged, and where many would-be solutions may today be lost.

A document gets a nod. A working prototype in front of a clinical or regulatory stakeholder gets a debate, which is what you actually wanted, while changing your mind is still nearly free.