Notes on building the AI-native web.
Page 13 of 17
Receiving webhooks without losing events
Someone else’s system calls yours, once, possibly twice, possibly while you’re deploying. Four rules that make it survivable.
Voice or text: which AI employee to hire first
Voice demos better and is harder in every way that matters. Text first is the boring, correct answer for most businesses.
Why your dashboard doesn’t get used
It answers questions nobody asked, at a moment nobody is looking. Four fixes, none of which is another chart.
When video helps on a marketing site
It is the heaviest thing you can add and occasionally the most persuasive. Three cases where it earns the weight.
When to add a queue
Queues fix slow requests and unreliable dependencies, and they introduce a whole new category of failure. Three signals that it’s time.
When the AI employee is better than your process
Sometimes the corrections stop coming because the system found a better way. Knowing the difference from a failure to notice is the job.
Brand systems for companies that don’t want to look like everyone else
Category conventions are a safe way to be forgotten. How to build a distinctive system that still works at speed.
The pre-launch checklist
Fourteen things worth checking before a site goes live, ordered by how much damage they do when missed.
How to introduce an AI employee to your team
The people who will make it work are the people it makes nervous. How you announce it decides which one you get.
How reviews show up in AI answers
Models read what customers wrote about you, not just your star rating. The text matters more than the number now.
What actually drives the cost of a website
Quotes for the same brief vary enormously, and it is rarely about design quality. Six variables explain most of the spread.
What happens when the AI employee is offline
Providers have outages and your integrations break. If the fallback is “nothing happens”, you have built a single point of failure into your front door.