Notes on building the AI-native web.
Page 10 of 17
What to do with the data your AI feature generates
Every interaction produces a record of what was asked, what was answered, and what a human did next. Most teams throw it away.
The form is where you lose them
Every field is a chance to leave. What to ask, what to cut, and why your qualification questions belong somewhere else.
What happens to your AI feature under load
Model calls are slow, rate-limited and metered. A traffic spike behaves very differently to one hitting ordinary code.
What an AI employee costs to build
The software is the cheap part. Four line items that dominate the build, none of which is engineering.
Do you need a mobile app or a good mobile site?
An app costs several times more to build and far more to keep alive. Three conditions justify it; nothing else does.
How to run an AI proof of concept in two weeks
Long enough to learn something real, short enough that cancelling isn’t a defeat. The shape that avoids permanent pilot purgatory.
Schema markup that actually changes what AI says about you
Most schema is decoration. A small subset genuinely disambiguates your brand for a model. Here’s which parts earn their place.
The AI employee onboarding checklist
Twelve things to have settled before it handles anything real. Most failures trace back to one of them being skipped.
The technical debt you should keep
Not all debt accrues interest. Paying down the wrong kind is how teams spend a quarter and change nothing a customer notices.
Measuring how customers feel about your AI employee
A thumbs-up button measures who bothered to click. Three signals that tell you what customers actually experienced.
The AI employee that uses your tools
Reading is safe. Writing to your systems is where an assistant becomes a colleague — and where the governance starts.
Is dark mode worth building?
For an app people live in, often yes. For a marketing site, it doubles your design surface to serve a preference most visitors won’t notice.