Success stories
They built it in a weekend. We made it survive Monday.
Ten apps built with Lovable, Replit, Base44, Claude Code, Cursor, Bolt and v0. What they looked like when they arrived, what was about to happen, and what changed in 48 hours. Names withheld; numbers real.
A product catalogue built on Replit now runs one of Europe's largest fashion e-commerce brands.
It started as a weekend prototype: one screen, one CSV, one very tired ops lead. Today it manages 311,000 size variants across six countries, eats 725,000 stock events a fortnight and survives a broken supplier feed at 3 a.m. Here is what changed between those two sentences.
Built with Replit · 14 min read · readiness 1.8 → 4.1
From 36 tests to 3,650 in 48 hours: the sales app whose data lived in the browser.
Investors loved the demo. The first serious client asked where the data was backed up. There was no data — every office, seller and lead was a JavaScript array. Two days later it ran on the client's own cloud with 3,650 tests and a load test at a year of traffic.
Built with Claude Code · 13 min read · readiness 2.0 → 3.4
560,000 customer profiles, two databases, zero duplicates, 50,000 emails a day.
The marketing team built their own customer data platform with Claude Code because the agency quote had six digits. It worked beautifully for 4,000 profiles. At 560,000 the nightly job took 45 minutes, segments read stale data and the API key defaulted to the string 'default_key'.
Built with Claude Code · 12 min read · readiness 2.1 → 3.9
Every customer could read every other customer's invoices. Forty of them were paying.
A Lovable-built invoicing tool with Supabase behind it. Beautiful UI, real revenue, Row Level Security switched off on the three tables that mattered and the service-role key sitting in the JavaScript bundle. The founder found out from us, not from a customer. That was luck.
Built with Lovable + Supabase · 11 min read · readiness 1.6 → 3.8
The clinic booking app that forgot every appointment when you refreshed the page.
An agency built it on Base44 for a chain of four clinics. Staff loved it for three weeks. Then a receptionist closed the tab. The data model existed only in the browser's memory; the 'database' was a demo template that reloaded on start.
Built with Base44 · 11 min read · readiness 1.4 → 3.7
The marketplace that ran in demo mode for eleven days and charged nobody.
Two founders, a Claude Code build, a launch on Product Hunt. Orders came in. Money did not. No environment variable was set in production, so the app started anyway — with demo defaults, the sandbox payment key and a feature flag that skipped invoicing. Nothing failed. That was the failure.
Built with Claude Code · 10 min read · readiness 2.2 → 3.9
The page that took fifteen seconds: passing an enterprise pilot with a Cursor-built app.
A leasing-broker tool built with Cursor won a pilot with a bank's partner network. The pilot brought 50,000 records. The dashboard went from 90 milliseconds to 15.8 seconds and memory quadrupled. The pilot manager gave them two weeks. We used two days.
Built with Cursor · 11 min read · readiness 2.3 → 3.8
Thirteen lost stock updates in one night, and the Monday-morning test.
An internal tool built on Bolt pushed stock levels to three storefronts. One night the storefront API returned 502 for four hours. The tool retried three times, gave up, and logged nothing. On Monday, 546 phantom units were on sale and nobody knew.
Built with Bolt · 11 min read · readiness 1.9 → 3.9
Promotions that started two hours late and barcodes that belonged to nobody.
A v0-built storefront admin for a brand selling in four countries. Two quiet bugs: date-times without a timezone read as UTC, and a supplier feed that put an internal counter into the barcode field and overwrote 193 real barcodes. The warehouse could not match its picking lists for a week.
Built with v0 + Next.js · 12 min read · readiness 2.0 → 3.8
Ship it switched off: the app whose business rules nobody had approved.
A non-technical founder built a staffing platform with Lovable and Claude Code over four months. It contained 64 business decisions — who gets assigned, what a hold costs, what happens on a duplicate — that she had never consciously made. The AI had. We shipped the app live in 48 hours with every unapproved rule switched off, and handed her a list.
Built with Lovable + Claude Code · 10 min read · readiness 2.0 → 3.5Send us what you have.
Discovery is $500: we look at your product and tell you, honestly, whether it is anywhere near production. If you go ahead, the $500 comes off the bill. 48 hours later, it's production-ready — on your own cloud, with the tests to prove it.