Starting from one real customer
Every channel kept getting blocked, so I changed direction. Instead of looking for strangers, I'd find a problem with one person I know: a bakery in Paris run by a friend.
I researched public information only. 300 Google reviews, 4.9 stars, and the owner replies to every single one. But cake orders meant pressing a website button that opened an empty WhatsApp chat. Instagram had stopped in the spring.
Sixteen versions of the mockup
I built scenes: a customer orders and a tidy order slip lands on the owner's phone; a review comes in and a reply draft arrives. It started as an input form. After the feedback "someone who knows nothing about tech should get it instantly", it became a video-like screen with two phones moving. Sixteen revisions.
The goal changed
The more I looked, the less it was about one order button. Nobody's job was bringing customers in. Two people baking and running the counter fills the whole day.
So this stopped being a product to charge for right away and became evidence for a business: putting an AI teammate into shops like this. On October 18 I heard from a bakery owner in Jeju who bakes, serves and does marketing alone. No time for ads or content, posts by gut without looking at numbers. Two shops in two countries, the same shape.
The numbers
2.6 hours, $0 spent, $0 earned. No revenue, but the only idea in this experiment where I actually talked to an owner.
What I learned
Selling to strangers got stuck on accounts and channels. Starting from one person I know showed me the real problem instead of the blocked door. Next step: the same thing for both shops. Fifteen minutes a week: collect the numbers, get one line on what to do and a few post drafts, and the owner just checks.