FigOCR
March 26, 2026
I've been hacking away on this project for five nights straight now, and I wanted to write down what's been driving it — partly to share, partly to think out loud.
Three or four years ago, I started building Screenshot to Layout as a Figma plugin. The premise is simple: take a screenshot — of a website, an app, a reference — and have it translated into a real Figma layout you can work with. Not a flattened image. An actual, editable structure.
Getting there required figuring out a lot of things in parallel: Figma OAuth, server key security, CORS issues, how to market a plugin, how to think like an indie hacker, how to be a better developer. It was a full education compressed into a side project.
The plugin almost died three times. What kept it alive was a grant from Figma's creator program. That money gave me enough runway to hire developers and take a real run at it. We connected to Supabase, built out an account system, implemented the initial OCR layer. Real progress.
But the core promise — taking a screenshot all the way through to a complete, structured layout — never fully materialized. The budget ran out before we got there. I'm grateful to everyone who tried to help make it happen, and I'm sorry it stalled where it did.
The tooling available right now changes what's actually possible. Claude's latest model supports up to one million tokens of context. Figma recently shipped write capabilities to their official MCP server, meaning you can now prompt your way into actual design changes inside a live Figma file. The infrastructure to build something genuinely useful is finally there.
The why behind Screenshot to Layout hasn't changed. I need to work in a system. My workday isn't five-minute experiments — it's eight hours of designing features, screens that reference components, components that live in a design system, a design system that connects to the codebase. I can't start from scratch every time. No serious designer can.
But even though my own work is structured, the input I get from stakeholders rarely is. I get screenshots, Excel files, data from all kinds of places — and I need to move all of it into Figma. There's no clean way to do that today. That gap is exactly what Screenshot to Layout is meant to close.
The output landing in Figma is non-negotiable for me. Figma is where production design actually happens. That's what 2.0 is about.
So that's what I've been spending my late nights on. More to come.