✨ Founder story

How Screenshot to Code turned a screenshot into $260K — and then went up for sale

Abi Raja posted a screenshot-to-code toy on a Friday in November 2023. Three years later it had earned $260,000. Then he listed it for sale and went back to a full-time job.

$260KAll-time revenue
$4.4KLast 30 days
281Subscriptions (Oct 2026)
$38KAsking price, Oct 2026

A Friday post

On 15 November 2023, Abi Raja — a founder in New York whose company WhimsyWorks also makes Pico, an AI app builder — posted a small tool he had built: paste a screenshot of a web page or an app, get back working front-end code. It was the kind of thing developers had wished for since the first AI code assistants appeared, and it arrived as a toy, not a company.

The next day it reached the front page of Hacker News. Three days later it was first on GitHub trending. The repository is open source under the MIT licence — anyone may run it, change it, or sell it — and that openness was the distribution: developers tried the code for free, starred it, and told each other. By 4 December 2023, three weeks after launch, the tool had made $1,000.

Design-to-code is an old idea with a long graveyard behind it — companies that promised to turn designs into code and died trying. What changed in late 2023 was that the models got good enough for the demo to be real: paste a screenshot, get back code that actually looks like the screenshot. Raja did not invent the idea. He shipped the version that worked, in public, on a Friday.

Open source as the funnel

The product is simple to describe: turn a screenshot, a design mockup, a screen recording or a text prompt into working front-end code in seconds — HTML and Tailwind, React, Vue, Bootstrap and more — then refine the result with plain-English follow-up prompts. The free code does the job; the hosted version at screenshottocode.com does it for you, selling credits instead of seats: one credit buys one generation or one edit, on plans at $15 and $40 a month.

That split is the whole business model. The repository is the marketing — every star, fork and Hacker News comment points back at the hosted version — and the hosted version is the business, charging the people who would rather pay than self-host. It is the classic open-source indie shape: give the code away, sell the convenience.

It also meant the product never needed a sales motion. No demos, no talk-to-sales button, no annual plan. A developer finds the repo, tries the hosted version, buys credits when the free ones run out. The 281 subscriptions on the books in October 2026 were mostly that: people who had tried it and kept coming back, month after month, for a tool they reach for in bursts rather than live in.

The numbers

The revenue record is unusually complete, because the Stripe account is connected to a public revenue page. The average month before October 2025 brought $8,585; the best single month was May 2026 at $6,478. By 14 October 2026 the account showed $259,933 in total revenue, of which about three quarters — $197,154 — had come before October 2025.

The recent direction is down, not up. September 2026 brought $4,727; the last 30 days brought $4,567, 46.8% less than the old average; monthly recurring revenue was $4,303 from 281 subscriptions, down 9.7% in 30 days. The product still earns a real living — roughly $4,400 a month — but the curve that once pointed up has flattened and started to slide.

That is the honest shape of most indie tools after year two: a strong launch, a long tail, then a slow fade as the novelty wears off and newer tools arrive. Screenshot to Code did not collapse; it decayed gracefully, still converting a few hundred subscribers a month into a few thousand dollars. The distinction between the two revenue numbers matters here: the $4,567 is cash in the last 30 days, the $4,303 is the recurring base underneath it — and both are roughly half what the product averaged in its first two years.

The sale

On 14 October 2026, the same day the revenue figures were published, Raja listed the business for sale at $38,000. The reason, in his own words, was that he is starting a full-time job. The asking price is about eight to nine months of current revenue — the standard multiple for a small tool with declining MRR and a single operator.

There is nothing shameful in that ending, and the listing is arguably the most interesting decision in the whole story. Most indie products do not get sold; they get abandoned — the domain lapses, the repo goes quiet, the Stripe account trickles to zero. Listing it turns three years of work into a clean exit and hands the tool to someone who wants to run it.

The lesson is narrower than the launch story. A public launch can buy you a year of growth; open source can buy you distribution; but neither buys you permanence. What they buy is the option to sell — a real business, with real revenue, that someone else might want. Raja built that option in a weekend in November 2023, and three years later he is cashing it in. The tool, meanwhile, is still live, still earning, and still waiting for its next owner.

The playbook

  1. Ship the toy in public — Post it on a Friday, let Hacker News and GitHub do the distribution: front page the next day, first on GitHub trending three days later.
  2. Open source as distribution — MIT licence: anyone can run, change or sell it. The free code is the marketing; the hosted version is the business.
  3. Sell credits, not seats — One credit buys one generation or one edit, on $15 and $40 plans. Usage-based pricing fits a tool people reach for in bursts.
  4. Know when to sell — With MRR drifting down and a full-time job starting, Raja listed it at $38,000 rather than letting it fade. An exit is a decision, not a failure.
🚀 Share your milestone View Screenshot to Code →