A few weeks ago, I launched a small niche job board focused entirely on jobs in football stadiums across the UK: stadiumjobs.co.uk
The interesting part is not the idea itself — niche job boards exist everywhere — but how absurdly cheap and fast it has become to build something polished using modern AI tooling and static site infrastructure.
This project became an experiment in how far you can get with:
- a cheap domain,
- a static site generator,
- GitHub hosting,
- and AI-assisted development.
Why UK Instead of England or Scotland?
At first, I considered targeting England specifically.
But from an SEO and branding perspective, “UK” turned out to be the cleaner choice.
There are several reasons:
- Many football clubs operate across England, Scotland, Wales, and Northern Ireland.
- “UK football jobs” has broader search intent than country-specific variations.
- A
.co.ukdomain immediately signals local relevance. - It avoids fragmenting the audience too early.
For a tiny niche project, simplicity matters more than perfect geographic precision.
Instead of creating separate brands for England and Scotland, I wanted one umbrella identity that could scale naturally.
Buying a Cheap .co.uk Domain
One underrated advantage of UK-focused projects is how inexpensive .co.uk domains still are.
You can often buy them for the price of a coffee.
That matters because niche projects are experiments. I don’t want infrastructure decisions that create pressure to “make the project succeed” immediately.
The lower the cost, the easier it becomes to test weird ideas.
A Vibe-Coded Jekyll Site Hosted on GitHub
The entire site is basically a vibe-coded Jekyll setup hosted on GitHub Pages.
No backend.
No database.
No Kubernetes.
No cloud complexity.
Just:
- Jekyll,
- Markdown,
- GitHub,
- and static hosting.
This setup has huge advantages:
- near-zero hosting cost,
- extremely fast performance,
- minimal maintenance,
- simple deployment pipeline,
- and almost no operational risk.
I can push updates directly via Git commits and have the site redeployed automatically.
For small SEO-driven projects, static generation still feels massively underrated in 2026.
Claude Opus 4.7 Dramatically Improved the Design
The biggest surprise was how much Claude Opus 4.7 improved the visual quality of the project.
Earlier AI-generated landing pages often had:
- awkward spacing,
- generic layouts,
- poor typography,
- and “AI-looking” structure.
But with Opus 4.7, the output became significantly more refined.
I used it to iterate on:
- page hierarchy,
- typography,
- responsive layouts,
- CTA positioning,
- card styling,
- and overall visual consistency.
Instead of generating one giant blob of HTML, the best workflow was conversational iteration:
- adjust spacing,
- simplify sections,
- reduce clutter,
- improve hierarchy,
- repeat.
The result feels much closer to a professionally designed indie startup page than previous generations of AI-assisted frontend work.
The Economics Are Kind of Incredible
This is probably the most fascinating part.
The total stack cost is almost nothing:
- cheap domain,
- free GitHub hosting,
- static site generation,
- no backend infrastructure,
- AI tools replacing large chunks of design and frontend iteration.
Ten years ago, building even a decent-looking niche job board required:
- a designer,
- a frontend developer,
- hosting setup,
- CMS configuration,
- and ongoing maintenance.
Now one person can prototype and launch something usable in a weekend.
The barrier between “idea” and “public product” has collapsed dramatically.
Final Thoughts
I don’t think the future belongs only to giant platforms.
AI tooling is making ultra-focused niche products viable again because the operational overhead is approaching zero.
That changes the economics completely.
You no longer need massive scale to justify building something useful for a small audience.
Sometimes a cheap domain, a static site, and a good AI model are enough to launch.
Top comments (0)