Why we built PageLens AI

Hi — I'm Anna. PageLens AI is built and run by me and my husband, a small two-person operation out of Surrey, England.

This is the story of how it started. It's also why he's not on this page.

A 20-year engineer who couldn't stop building things

My husband has been writing software since the early 2000s — long enough to have shipped production code in ASP, .NET, Java, and a parade of other languages he'd rather not list. He's built databases, web services, desktop apps, and web apps. The kind of engineer who knows the whole stack because he was there when half of it was being invented.

For five of those years he specialised in performance — the unglamorous, deep-end side of the job where you fly to other countries to figure out why a Fortune-500 retailer's checkout falls over on a Tuesday afternoon. He still works full-time as a senior engineer at a major tech platform today.

Which is exactly why his name isn't anywhere on this site. PageLens AIis a side project, his employer doesn't allow them, and we like having a roof. Mine's OK with it. Hi, it's me, Anna, the registered owner.

The moment a side project became necessary

Around the start of 2024 he started using AI agents to build small products in his spare time. Cursor, Claude, the whole stack. They worked — sites were shipping in a weekend that would have taken a month to build by hand. He was thrilled.

They were also shipping broken.

Lazy security headers. Form inputs without labels. Hero images served at 2.4MB. Hallucinated meta tags. Footer links pointing to /about#team with no #teamanywhere on the page. The kind of issues that don't show up in a happy-path demo but absolutely show up when a real user — or a real attacker — visits.

He'd built the AI scaffolding for himself. Now he needed an audit scaffold to keep it honest.

So he built one. It crawls every page in a real headless browser, screenshots them, reads the rendered HTML and the response headers, runs deterministic checks for the boring stuff (CSP, alt text, broken anchors, security headers) and a vision-capable AI for the interesting stuff (does the hero actually communicate what the site does? Is the CTA below the fold on mobile? Does the brand voice survive the journey?).

Originally just for his own sites. Then for one of mine. Then a friend's. Then someone in a Facebook group asked if they could pay for it.

Why now

Roughly 550,000 apps were deployed to the major no-code and AI-coding platforms last year. A growing share of them are written largely by AI. A growing share of those ship with security headers a 2007-era PHP site would be embarrassed by.

The traditional answer is to hire a consultant. £5,000, six weeks, a 60-page PDF nobody reads. That's not a serious option for an indie founder testing whether their idea has legs, or a small agency triaging six client sites at once, or a junior PM trying to understand why their conversion rate dropped.

We charge from $1 because the AI is doing the work, not a $200/hour human. We make our money on volume, on customers who come back monthly to re-scan, and on people who level up to the bigger crawl tiers as their sites grow. The economics work without us having to pretend they don't.

Who's actually behind it

Day-to-day:

• Anna Moore(me) — registered owner, customer support, marketing, and the person you'll get if you email hello@pagelensai.com.
• My husband — full-stack engineering, the audit engine, the AI prompts, the infrastructure. Quiet on the public side, for the reasons above.

That's it. No investors. No growth-at-all-costs targets. We built this because we needed it ourselves, and because charging a fair price for a useful tool turns out to be a perfectly reasonable way to make money on the side.

The most recent thing we shipped is the part of the product we're proudest of: a verified-domain badge that turns your score into something you can actually publish. Prove you own the domain, embed a signed PageLens AI badge in your footer, and we host the public trust page anyone can visit to confirm the score. Other audit tools generate PDFs and call it a day. We thought the score should be showable— the way a code-coverage badge or an SOC 2 stamp is showable — and so far the people who've embedded one tend to keep running scans.

If you're shipping AI-generated code and the back of your mind keeps whispering "is this actually any good?" — we built it for you, too.

— Anna