I watched a reel recently where Charlie Weavers-Wright, a venture capitalist, was explaining why investors have stopped chasing AI application layer startups. What he said mapped almost perfectly onto the thing we’ve spent the last few months building.

The LLM wrapper problem

The argument he was making is this: the tools that do little more than put a conversational front-end on top of a large language model are in trouble. Not because they’re badly built, but because the LLMs themselves are getting good enough to make the wrapper redundant. Why pay for a tool that summarises your emails when the email client will do it natively? Why pay for an AI writing assistant when the word processor already has one?

So where is the smart money going instead? Two places. First: the infrastructure layer – the unglamorous, essential, mission-critical plumbing that makes AI systems work properly at scale. Data pipelines. Evaluation frameworks. Fine-tuning infrastructure. Tooling. The boring stuff that never makes a headline but without which none of the exciting stuff functions.

Second: founders who know something deeply. Not people who spotted an AI trend and built a product around it, but people with genuine, hard-won, domain-specific expertise. People who understand the pain points from the inside, because they’ve lived them.

SUMMARY

The products most likely to survive the next wave of AI capability are the ones solving real, specific problems that take years of experience to even recognise.

When I heard all that framing, I thought – that’s absolutely SiteVitals.

What SiteVitals actually is

SiteVitals is a website monitoring platform that watches websites the way a vigilant, full time developer (with an empty to-do list) would, and alerts you when something changes, breaks, expires, or degrades. Uptime. SSL certificates. Domain health. Performance and Core Web Vitals. SEO fundamentals. Security signals. Third-party script changes. Content drift. All of it, continuously, with enough context to act on rather than just a notification that something’s wrong.

It’s not an AI product in the chat-interface sense. It uses intelligent algorithms – a Weighted Change Index, semantic keyword drift detection, dependency fingerprinting – to do something that requires genuine technical depth to design well. It is, in the investor’s framing, closer to infrastructure than application. It’s the layer underneath the website that tells you whether the website is actually doing what it should.

Why this matters now?

With AI-powered search changing how businesses are discovered online, the technical health of a website has never been more consequential. Agentic AI tools increasingly decide which sites to surface and recommend – and they rely on sites being technically sound. Monitoring isn’t a nice-to-have, it’s the baseline for visibility.

The domain expertise angle

Tom and I have been running our web agency, 18a Productions, since 2007. Between us and our team, we’ve built and supported websites for everyone from solo traders to organisations with sites handling 2.4 million visitors a day. We’ve been the outsourced technical backbone for design agencies, helped build digital departments, worked with the NHS, BBC Two, Aardman, national charities and growing brands.

More relevant to SiteVitals: we’ve spent over two decades watching what actually goes wrong with websites in the real world. Not the dramatic things. The subtle things. The domain renewal email that went to a former employee’s inbox. The homepage that tripled its load time because someone uploaded a photo straight from their phone. The SSL certificate that didn’t auto-renew and sent browsers into a “Not Secure” spiral before anyone noticed. The Google ranking that slid for six weeks before anyone connected the dip to the plugin update.

These are not hypothetical scenarios we’ve constructed to make the product sound useful. They are things we have seen – repeatedly, across every size of business, across over twenty years of client work. Charlie Weavers-Wright’s second criteria – founders who know their domain deeply, who understand the pain points because they’ve lived them – that’s us. Genuinely and specifically.

Starting out when we did, you didn’t specialise. To build and launch a website in the early 2000s, you had to understand all of it – front end, back end, DNS, hosting, domains, performance, security. That full-stack perspective is still baked into everything we build.

Lisa Freeman

Director

The infrastructure parallel

The investor’s point about infrastructure is worth unpacking a bit further, because I think it applies more broadly than just AI companies. What he was really saying is: the tools that sit underneath the things people care about – the monitoring, the pipelines, the evaluation, the checks – are more defensible than the tools that sit on top of them. Because the underlying problem doesn’t go away just because the interface changes.

Websites will keep needing to be watched. Certificates will keep expiring. Scripts will keep breaking. Content will keep drifting. Rankings will keep fluctuating. None of that is going away because LLMs get better – if anything, the consequence of getting it wrong is increasing, because AI search is raising the stakes on technical quality. SiteVitals sits in that layer. It’s not dependent on any particular AI wave. It’s solving a permanent, specific, expensive problem for businesses that don’t have a developer watching their site around the clock.

That feels like a pretty good place to be building something in this time of massive change to our industry.

We didn’t build it because it was fashionable

One more thing the investor said that stuck with me: the products he’s most sceptical of are the ones that were built because someone spotted an opportunity in the AI hype cycle. The ones he believes in are built because a founder couldn’t find something that did the job, and decided to build it themselves.

SiteVitals was my idea, and it came directly from years of watching our clients encounter the same avoidable problems. I wanted a tool that would catch everything before it became a crisis —- something that would be the professional eye on their site between agency visits. Tom then took that idea and built something considerably better than I’d imagined, as he tends to do.

We’re not looking for VC investment, for the record. We’re a small, agile, self-funded team and that’s exactly how we want to stay. But it’s just encouraging to hear smart investors articulate, in their own language, exactly why the thing we’ve built is a good idea.

You can explore SiteVitals at sitevitals.co.uk — uptime monitoring is free forever, and full monitoring starts at £19 a month. No enterprise pricing. No venture-backed sales team. Just the tool.