We Built an AI Canvas for Everyone. Then We Threw It Away.

October 2025. Two founders, one idea: an AI canvas where you could drag, drop, and connect every major AI model in one place — image generation, video, editing, all of it, in one visual workspace.

We built it in two months flat. Late nights, no sleep, pure obsession. The kind of building where you forget what day it is.

We pitched it. People loved it. A waitlist started filling up. For a few weeks, we felt unstoppable — that specific high of watching something you made from nothing start to get traction.

Then February happened.

We sat down and asked ourselves one question: "Who is this actually for?"

Silence.

Agencies could use it. So could freelancers. So could random creatives messing around on a Tuesday. It did everything, for everyone. And that's exactly the problem — a product for everyone is a product for no one. Nobody could describe it in one sentence. Nobody was desperate enough to pay for it. People loved the demo and then went back to their lives.

We had a working product and a broken business.

So we did something that scared us: we threw it away.

March: Pick One, Go Deep

New rule — pick one industry, go deep, not wide. We landed on jewelry photography. It wasn't a random pick — jewelry has been in the background of my life since childhood, hearing about the trade from family. But it wasn't just sentiment either — it's a vertical with a real, specific, expensive problem: getting product photography that actually does the product justice is slow and costly, and most small and mid-size jewelers can't afford a proper studio setup for every SKU.

Fifteen days. Complete rebuild. From scratch. Same underlying AI capabilities, but this time built around one workflow, one customer, one problem — instead of a general-purpose canvas trying to be everything to everyone.

May: More Ground, Less Laptop

We didn't touch our laptops as much as we touched the ground. Real jewelers, real shops, real feedback. We listened more than we built. That part was uncomfortable in a different way than the late nights were — it's easier to disappear into code than to stand in a shop and hear someone tell you your assumptions are wrong.

Today

We're not chasing everyone anymore. We're chasing the right people, hard. The product is narrower than it used to be. It's also, for the first time, actually a business — not just a demo people admire.

If you're building something right now and it feels like it's "for everyone" — it's probably for no one.

Pick one. Go deep. That's when things start moving.

We're building Grafx — an AI-powered product photography platform for jewelers. Turn a single photo into studio-quality shots — model shoots, styled stills, multi-angle images — in minutes instead of days, at a fraction of traditional shoot costs. Check it out at grafx.app.