What Went Down, and What Went Up
Twelve posts, sixty projects, one year. Looking back at the tunnels and the take-offs together, a few things were true every time.
Build it to find out
Almost nothing this year was settled by thinking. The small-app questions, the Headless bet, the AI visibility numbers, the pitches: each was answered by building a small version and looking at it. When building is quick, the build is the research.
Every tunnel fed a take-off
The parked job-tracking app became a client's system. The App Market work led straight into Wix Headless. A game for my son became a showcase. The estimated AI scores became measured ones. Almost every tunnel fed something that took off.
Write down what you learn
Every time something went wrong twice, it became a rule: where documents live, how releases are checked, what a pitch must never do, what "done" means. A year of rules is now the most valuable thing the studio owns.
And don't go alone
I built all of this with Claude, and the way we work changed through the year. Here are the habits I'd pass on to anyone building with AI:
- Decide and guard in one line. "Option A. Before you start, check your assumptions."
- Quote, then correct. Paste the exact sentence you disagree with and answer it.
- Show, don't describe. A link to a site you like beats a paragraph of adjectives.
- Write the real user's test. "Try it just like my client would."
- Turn complaints into rules. If you've said it twice, write it down once.
Into 2027
Some tunnels stay open. A version of the studio's brain for other small businesses. Pitches that turn into paid work more often. A few new apps for Wix Studio that are still on the bench. And more take-offs on Wix Headless, which is where I expect much of next year's work to be.
Thanks for reading all twelve. Here's to 2027.