5 companies that shipped real products with vibe coding

A MacBook with lines of code on its screen on a busy desk

A UK doctor was quoted £75,000 to £100,000 to build a healthcare platform. He built it himself in four days for £165. That is the number that earns vibe coding a serious look.

Vibe coding, the practice of describing software in plain English and letting AI handle the code, testing, and deployment, was coined as a term in early 2025. Since then, five businesses ranging from a solo GP practice to a 3,500 person engineering team have run real experiments with real outcomes. Here is what they found.

Booking.com: 30% productivity lift after training clicks

Booking.com rolled out AI coding assistants to its entire 3,500 person engineering team in early 2025. Adoption was initially slow. According to Bruno Passos, group product manager for gen AI, things only shifted after the company invested in structured training on how to use the tools effectively.

Once developers understood how to get results, measured output of finished, reviewed work climbed 30 percent. Job satisfaction went up alongside productivity. The lesson Passos draws is clear: tools alone do not change behavior. Training and a culture that rewards experimentation do.

3D rendered ai text on dark digital background

Zapier: a growth marketer built his own dashboard

Matt Brown, a growth marketing manager at Zapier, needed to track data from influencer partnerships. Rather than filing a ticket with an engineering team, he built the tool himself using Claude and Zapier’s own systems.

The custom internal dashboard generates data architecture, JavaScript, and HTML on the fly. It also automatically spots and fixes bugs. The result: surface level stories and insights that were not visible from looking at raw data. The dashboard now runs in Zapier’s weekly team briefings.

Adidas: a rough first pilot, then a 50% jump in coding time

Adidas reported that its first vibe coding pilot went badly. Ninety percent of participating developers described the tools as a complete waste of time and said they hated using them. That is worth noting. Not every team adopts these tools smoothly on the first attempt.

A second pilot produced very different results. Developers reported feeling 20 to 25 percent more effective. More significantly, vibe coding increased what the team called happy time, meaning time actually spent writing code rather than on administrative work, by 50 percent.

laptop screen displaying colorful code

Northern Health: 16 features in 4 days for £165

This is the case study that gets passed around for good reason. UK GP Dr Fahim Hussaun needed a remote healthcare platform for his Northern Health practice. Quotes from traditional developers came in at £75,000 to £100,000, with a 12 to 18 month build timeline.

Using the vibe coding platform Replit, Dr Hussaun built 16 features of the My Doctor platform in four days. The feature list included appointment booking, prescription requests, a subscription payment system, and integration with Fitbit tracking devices. The reported development cost was £165.

For a small practice weighing a six-figure capital risk against an 18 month wait for revenue, the comparison is not subtle.

Leatherman: 250 internal projects shipped by non-developers

Leatherman, the physical tools company, extended vibe coding access beyond its engineering team to staff across manufacturing, marketing, finance, HR, and operations. The instruction was straightforward: build solutions to the problems you face every day.

The result was more than 250 projects going into production, including apps, dashboards, and automations. The people building them were not engineers. They were the workers who understood the problems best.

What the pattern shows

These five examples came from different industries, team sizes, and starting conditions. But two things showed up consistently:

  • Training determines outcome. Both Booking.com and Adidas saw poor early results that improved significantly after developers learned how to use the tools. Dropping software in front of a team without support does not work.
  • Non-technical operators can ship real software. Matt Brown built a dashboard that runs weekly briefings. Dr Hussaun built a 16 feature healthcare platform. Leatherman’s manufacturing staff built production automations. None of them needed to hire a developer.

The ceiling on what non-technical operators can build is still being discovered. But the floor, based on these five cases, is higher than most organizations currently assume.

Stay on top of AI & Automation with BizStack Newsletter
BizStack  —  Entrepreneur’s Business Stack
Logo