Petzo is a pet-care marketplace in Indonesia: clinics, pet hotels, and pet food in one place. I had already spent about a week designing it in Figma when the founder's investor said the thing every designer eventually hears, that a picture of a product cannot be used, pitched, or tested. So I put Figma down and built it, alone, with AI. Catalog and checkout in a week. The full MVP in under a month.
The fastest MVP isn't the one you build. It's the one where you wire together what already exists and build only the ten percent that's actually yours.
An early-stage founder needed something an investor and a pet shop could both react to. My Figma file was not it. You cannot hand a static screen to a shop owner at an event and learn whether they would sign up, and you cannot put a prototype in front of an investor and have it count as a product. That week of design was not wasted, but it had reached the limit of what a picture can answer.
Designer, PM, and builder. Nobody else was going to scope this, so I owned what went in and what stayed out, then executed the whole product on AI with barely a Figma file in sight. It was the first time I took a product all the way through that way, and the founder gave me the room to try it.
The temptation with a marketplace MVP is to build the marketplace. I scoped it the other way, by asking which parts of this product were genuinely ours:
- Payments: not ours. Redirect to a payment gateway rather than build a checkout stack.
- Notifications: not ours. Order confirmation over WhatsApp, where Indonesian shop owners and customers already live.
- The catalog, the booking, the admin: ours. That is roughly ten percent of a marketplace, and it is the only part worth spending a month on.
That decision is what made the timeline possible. The foundation, a catalog across categories plus the checkout flow, took about a week. The full MVP, including an admin dashboard so the shop could manage order status, landed in under a month.
Petzo launched at Indonesia's biggest pet event, and the founder used the live product to talk to prospective investors and partners. Pet-shop onboarding went from 3 to 11, off a base of almost nothing, and the unlock was simply having something a shop owner could hold instead of imagine. Then the loop got interesting. Partners started bringing real operational detail, one shop already ran its own scheduling, another exposed a gap between two services we had treated as one, and because the build was fast we could turn that input into an updated product to show them the next day. Business development stopped being a demo and became an iteration cycle.
AI did not make me a faster designer. It moved me from designing screens to validating the business, which is a different job with a different scarce skill. The production gap is closing; the judgment gap is not. What still had to come from me was deciding what not to build, reading a partner's operational reality correctly, and knowing which ten percent was worth owning. That is also where design and engineering stop being two rooms: when the person making the call can also ship the thing, the argument gets settled by the product rather than by a handoff.