What is vibe coding? A plain-English explainer (2026)
By Peter Peart · 3 min read
If you've been anywhere near tech Twitter or LinkedIn in the last twelve months, you've seen the phrase. Vibe coding. Half the people using it sound like they've unlocked a cheat code; the other half are rolling their eyes. So what is it actually?
In one sentence: vibe coding is building software by describing what you want to an AI in plain English, then letting the AI do the typing. You prompt, it builds, you prompt again, it refines. Tools like Lovable, Cursor, Bolt and v0 all sit somewhere on that spectrum.
That's the short version. The slightly longer version — and the bit that matters if you're trying to decide whether to bet your weekend (or your business) on it — is more interesting.
The thing it isn't
It is not "the AI does everything and you sit back." Anyone telling you that has either never shipped a real app, or is about to learn an expensive lesson in front of paying customers.
Vibe coding shifts the bottleneck. The typing gets faster — dramatically faster. But the bottleneck is no longer typing. The bottleneck is thinking. What do you actually want? Who's it for? What happens at the edges? What does the data look like in six months? Those questions don't go away just because the AI can spin up a beautiful interface in twenty seconds.
If anything, the questions matter more. A traditional developer is slow enough that a vague brief gets corrected by reality before too much damage is done. An AI is fast enough to confidently build the wrong thing in an afternoon.
What it's actually good at
I've shipped five-plus production apps this way for paying clients — real users, real data, real money — and the pattern is consistent. Vibe coding is brilliant for:
- Lean, focused apps that solve one specific problem. Property management, maintenance ticketing, simple booking systems, internal tools, niche SaaS — the stuff that runs on WhatsApp groups and spreadsheets in real businesses.
- Getting from idea to a thing-you-can-show in days, not months. That changes what's economically viable to build.
- Iterating fast on real feedback. Because changes are cheap, you stop being precious about decisions.
What it's not for: the next Facebook. Apps that need to support hundreds of thousands of concurrent users, or genuinely complex infrastructure, still want a proper engineering team. But you almost certainly don't need to build Facebook. You need to build something people will pay for.
The unsexy opportunity
The most interesting thing about vibe coding in 2026 isn't the consumer apps. It's that suddenly it's economically viable to build proper software for boring industries. Property managers. Cleaning companies. Small manufacturers. Plumbers with twelve vans. These sectors have been ignored by software for decades because the market was too small to justify a team of engineers — and now it isn't.
A lean SaaS product priced at a few hundred pounds a year, solving one painful problem, is transformative for them and a perfectly good business for you.
The 70-app problem
Quick warning. There's a particular flavour of LinkedIn post that goes "I built 70 apps this year with Lovable." That number tells you nothing. One app solving a real problem for fifty paying customers at fifty pounds a month is worth infinitely more than seventy demos nobody uses.
Build less. Think more. Ship the one.
So is it right for you?
If you're a founder, a domain expert, or a consultant with a real customer in mind for a real piece of software — yes, this is probably the most leveraged way to build right now.
If you want to be a "vibe coder" who collects framework merch — that's not what this is.
The full first chapter of the Field Guide goes deeper on this: the 70-app problem, the unsexy-industries thesis, where vibe coding fits in your stack, and the cases where I genuinely tell clients to hire a traditional dev team instead. It's the chapter I wish someone had handed me when I started.