Thinking,in ink.

Patients search by symptom, condition, and body part. Hospital sites answer in the language of departments, so every marketing dollar drives traffic into a vocabulary wall the navigation built.

Every university homepage is cluttered in the same way, and the reason is not design. The space is divided by internal seniority, not visitor need, and no layout fixes a negotiation.

AI can generate a competent website in minutes, which is exactly why competence no longer sets work apart. What stands out now is whether the experience makes anyone feel something.

When a product feels broken, the instinct is to redesign the surface. Usually the problem is underneath, in how the thing is organized. The fix is structure, not paint.

Forward-thinking teams are trading static, text-heavy PRDs for working, clickable prototypes built fast with AI. The spec is becoming something you use, not something you read.

Most products do not fail because they lack features. They fail because they ask people to think too hard about the wrong things. A working method for spending attention deliberately.

A patient books care in one continuous motion across three systems owned by three teams. The worst failures live in the seams between them, and nobody is accountable for a seam.

Structural work pays off on a lag, as search re-indexes and habits re-form long after the quarterly review has rendered its verdict. Negotiate the measurement window before launch.

Everyone knows which pages are dead. They stay up because publishing is rewarded and pruning is punished, and making subtraction safe is a leadership deliverable, not another audit.

Months of recommending the same navigation change moved nobody; one tree test ended the argument in a single meeting. The difference was research designed so the losing side could concede safely.

Reviews approve finished-looking work because finish used to mean effort. AI made finish free, and the instinct that reads polish as rigor is now approving hollow work it cannot feel.

Design decisions used to be protected by the cost of producing an alternative, and AI removed that cost. If the reasoning was never written down, it is now up for grabs.

A junior on my team turned a rough brief into a polished prototype before lunch, and it worried me. The production drudgery AI just removed was where judgment formed, and that training now has to be rebuilt on purpose.