You can now type a few sentences into an AI website builder and get a slick, modern site back in an afternoon. Lovable, Bolt, v0, Replit and the rest are genuinely impressive. For a quick landing page behind a Google Ads campaign, they can do the job.
But a website that looks finished and a website that gets found are two different things. A lot of these sites are built in a way that search engines and AI assistants find hard to read. Most are also one page, which means they can only ever compete for a handful of searches.
Here’s the plain English version, then the detail, with links to the sources so you can check every claim yourself.
The short version
- Many AI-built sites are JavaScript apps. When a visitor arrives, the server sends an almost empty page, and the browser builds the content using code. People see a full website. A program that doesn’t run that code sees a blank page.
- Google can usually run the code, but it’s an extra step. Google puts these pages in a queue to be rendered. It recommends server-side rendering or pre-rendering anyway, because “not all bots can run JavaScript”.
- The big AI crawlers don’t run it at all. Vercel, the company that hosts many of these sites, studied its own traffic and found that none of the major AI crawlers, including those from OpenAI, Anthropic and Perplexity, rendered JavaScript.
- One page means one shot. A single page has one title and one address. You can’t be the best answer for “house removals Blanchardstown”, “office removals Dublin” and “packing tips” from the same page.
- You often can’t edit it yourself. A new service page, a price change or a blog post usually means going back to the AI tool, or paying a developer who understands the code.
How a normal website and an AI-built app differ
On a traditional website, like most WordPress sites, the server sends back a complete page. The words, headings and links are all in the first thing that arrives. Anything that reads that page, whether it’s Google, ChatGPT, Facebook’s link preview or an SEO tool, gets the full text straight away.
Many AI builders produce a single page app, usually in React. The server sends a small shell of a page and a bundle of JavaScript. The visitor’s browser then runs that code to draw the text, images and menus. It feels fast and app-like for people. For anything that reads the raw page without running the code, there’s very little there.
Not every AI builder works this way, and some have improved. We cover that below, because it matters.
What Google sees
Google does run JavaScript, so a JavaScript site isn’t invisible to Google. But Google describes it as a separate stage. Pages are crawled, then queued for rendering, then indexed. In Google’s words, a page “may stay on this queue for a few seconds, but it can take longer than that.”
Google’s own guidance still says server-side rendering or pre-rendering “is still a great idea”, because it makes your site faster for people and crawlers, and because not all bots can run JavaScript.
There are other traps Google documents for single page apps:
- Fake “page found” answers. Google notes that single page apps often answer every address with a “200 OK” status, even for pages that don’t exist, instead of a proper 404. Google calls these soft 404s.
- Addresses Google can’t follow. Google says you can’t rely on URL fragments (the part after a #) and recommends using proper addresses for each piece of content.
- The same title everywhere. Google asks for a distinct, descriptive title on every page and says to avoid repeated or boilerplate titles. One page apps often carry the same title on every screen.
What ChatGPT, Claude and Perplexity see
This is the part most business owners haven’t heard. In December 2024, Vercel published a study of AI crawler traffic across its network. Its conclusion: “none of the major AI crawlers currently render JavaScript.”
That list included OpenAI’s GPTBot, OAI-SearchBot and ChatGPT-User, Anthropic’s ClaudeBot, and PerplexityBot. ChatGPT’s and Claude’s crawlers did download JavaScript files, but Vercel found “they don’t execute them.” The exceptions were Google’s Gemini, which uses Googlebot, and Apple’s crawler, which renders JavaScript like Googlebot does.
These crawlers aren’t small. In that month Vercel counted about 569 million fetches from GPTBot and 370 million from Claude, roughly 28% of Googlebot’s volume when you add Apple and Perplexity in.
OpenAI is clear about why this matters for business. Its OAI-SearchBot crawler is what surfaces websites in ChatGPT’s search answers. If your text only exists after JavaScript runs, a crawler that doesn’t run it can’t quote you, can’t recommend you and can’t send you the customer.
Vercel’s own advice is to server-render the important content: articles, product details, metadata and navigation.
To be fair to Lovable and the others
Some builders have been working on this, and it’s worth being accurate:
- Lovable says new apps since 13 May 2026 are built with server-side rendering, so “every request returns fully rendered HTML, for humans and crawlers.” Older Lovable sites built on React and Vite get pre-rendered pages when a verified crawler arrives.
- Lovable’s own documentation lists the limits. Content that loads late “may be missing”. “Unverified bots and link checkers see the regular single-page app.” And most social platforms “do not run JavaScript”, so each page needs its own preview tags.
- Sites exported from these tools and hosted elsewhere, such as on Vercel or Netlify, don’t automatically get Lovable’s pre-rendering. They’re plain JavaScript apps unless someone set up server rendering.
So the honest answer is: it depends how the site was built and where it’s hosted. That’s exactly the problem. Most business owners have no way of knowing which one they’ve got.
The bigger problem: one page sites hit a ceiling
Even a perfectly rendered one page site has limits. Google finds pages mainly through links from other pages. It ranks individual pages for individual searches. One page gives you one title, one address and one main topic.
Think about how people actually search for a local service:
- the service plus their town (“removals Blanchardstown”)
- a specific job (“piano removals Dublin”)
- a question (“how much does a house move cost in Dublin”)
- a comparison (“man with a van vs removals company”)
Each of those deserves its own page, with its own title, its own address and real local detail. Google’s SEO Starter Guide recommends grouping similar pages in folders and using words in addresses that are useful to people, like /house-removals/dublin/blanchardstown/. You can’t do any of that on a single page.
That’s why these sites can work behind an ad, where you pay for every visitor, and then struggle to grow on their own.
A client example
A client came to us recently with one of these sites. It was built with an AI website builder: cheap, cheerful and quick to get live. It looked modern, and he thought he’d got a good deal. On the surface, he had.
Under the bonnet it was a different story. The site was a JavaScript app, so the raw page had only a few words of text before the code ran. Every page shared the same title. Made-up addresses came back as “page found”. And there was nowhere to add location pages or blog posts without going back to a developer.
His Google Ads were doing all the heavy lifting. Organic search wasn’t helping at all.
We’re rebuilding it on WordPress using Sistaiom, our own system for building WordPress sites, and keeping the look he liked. Each service and area gets its own page, linked together, with FAQs, photos and price guides. There’s a blog for practical tips, and he can edit everything himself.
Why we build on WordPress
We don’t build on WordPress because it’s old and safe. We build on it because it solves the problems above out of the box:
- Every page arrives complete. Google, ChatGPT, Claude, Perplexity and Facebook all get the full text on the first request.
- Unlimited pages and posts. Add a new town, service or guide in minutes, each with its own title, description and address.
- Proper SEO controls. Titles, descriptions, sitemaps, redirects, schema and real 404s are standard.
- You can edit it. With Elementor Pro and ACF Pro, changing a price or adding a photo doesn’t need a ticket to a developer.
- You own it. Your hosting, your files, your domain. No tool subscription holding the site hostage.
None of this means AI is the enemy. We use AI every day in our own work. The point is that the platform underneath matters. AI is great at writing and designing. It doesn’t replace a site structure built for search.
How to check your own site in 2 minutes
- View the raw page. On a computer, open your homepage, right click and choose “View page source”. Search for a sentence you can see on the page. If it isn’t there, crawlers that don’t run JavaScript can’t see it either.
- Test a made-up address. Add
/this-page-does-not-existto your domain. You should get a proper “page not found”. If you get your homepage, that’s a soft 404. - Check your titles. Open two or three different pages and look at the browser tab. If they all say the same thing, Google sees the same.
- Count your pages. In Google, search
site:yourdomain.ie. If you only see one or two results, you’re competing with one page. - Ask who can edit it. If changing a price means going back to the AI tool or a developer, you’re locked in.
Not sure what you’re looking at? Book a free 15 minute call and we’ll check it with you and tell you straight whether it needs rebuilding.
FAQ
Can Google read my AI-built website?
Usually, yes. Google runs JavaScript, but it renders pages as a separate step that can take time. Google still recommends server-side rendering or pre-rendering, and single page apps have other issues like soft 404s and repeated titles.
Can ChatGPT read my AI-built website?
It depends how it’s built. In Vercel’s 2024 study, none of the major AI crawlers, including OpenAI’s, rendered JavaScript. If your content only appears after JavaScript runs, they see very little. Sites with server-side rendering or pre-rendering send the full text.
Is Lovable bad for SEO?
Not necessarily. Lovable now builds new apps with server-side rendering and pre-renders older ones for verified crawlers. The bigger limits are usually a one page structure and needing the tool or a developer for every change.
Should I rebuild my site on WordPress?
If your site is one page, you want to rank for more than a couple of searches, or you can’t edit it yourself, it’s worth a conversation. If it’s a landing page for ads and that’s all it needs to do, it may be fine as it is.
Sources and further reading
- The rise of the AI crawler, Vercel (December 2024)
- Understand JavaScript SEO basics, Google Search Central
- Fix Search-related JavaScript problems, Google Search Central
- Influencing title links in Google Search, Google Search Central
- SEO Starter Guide, Google Search Central
- Overview of OpenAI crawlers, OpenAI
- Does Anthropic crawl data from the web, and how can site owners block the crawler?, Claude Help Center
- Optimize your app for SEO and AI search, Lovable documentation