That’s the real tension nobody’s addressing. A website built with AI might look polished and run smoothly. Search engines might have a completely different opinion about how to treat it. Google’s own search team recently dug into this exact problem on their podcast, and what they found matters whether you’re a developer, a business owner, or someone considering vibe coding for your next project.
AI Can Build Sites, But SEO Requires Actual Direction
Google Search Relations team members John Mueller and Martin Splitt recently discussed vibe coding websites on a podcast. Both found that AI coding tools could produce functional sites quickly. But getting SEO right still required specific technical direction, the same kind you’d give a human developer. That’s not a bug. That’s just how it works.
Mueller compared the experience to working with a developer who doesn’t specialize in search. Imagine telling them “add some SEO to it.” Most would ask: What do you mean by that? Should I sprinkle in meta tags? Add structured data? The instructions are too vague.
Here’s what Mueller actually said on the podcast: “You can always tell the AI system, now add some SEO to it. But how that works out is if you go to a developer and add some SEO and it’s like, what do you mean. Sprinkle some meta tags and add some structured data.”
The problem isn’t unique to AI. Vague instructions produce vague results whether you’re talking to a human or a machine. Mueller got better outcomes by being specific from the start. He told the system exactly what he wanted: the domain name, canonical setup, sitemap files, and a robots.txt structure.
He also checked whether pages used reasonable HTML and linked properly. He set up pre-publish checks to verify that URLs returned content and that JavaScript files weren’t blocked by robots.txt. That level of detail makes the difference.
The Tools They Actually Used and What They Found
Mueller built test websites to see how Googlebot handles requests. He deployed them to Firebase hosting using Hugo as a static site generator, with GitHub for version control. He’s tested different AI tools too. He started with VS Code and Copilot, then switched to command line tools like Claude Code and Gemini CLI.
Splitt took a different approach. He tried Google AI Studio to build a client-side tool with JavaScript. The output looked good. It read like a standard Next.js application. But then he hit a wall. The AI kept using a library he didn’t want, and no amount of asking would make it stop.
Splitt described it this way: “I asked it for a half an hour. I tried to make it not do what it wanted to do, and want to do what I wanted to do. And that was weird.” That’s the real experience. AI isn’t being stubborn on purpose. It’s just making its own assumptions about what works.
The Technical Knowledge Problem Nobody Wants to Admit
Here’s the uncomfortable truth both Mueller and Splitt acknowledged: vibe coding promises you don’t need to know how to code. But technical understanding actually helps at every stage.
Mueller pointed out that knowing what kind of site generator you want and how to structure pre-publish checks produces dramatically better results. Without that background, the AI will make assumptions. It might pick a static site generator, go heavy on JavaScript, or choose a full CMS with a database backend.
Mueller explained: “All of these are reasonable assumptions where if you talk to a developer they will also make these assumptions. But if you just tell the AI system like I want a website, then it will pick one.” The AI isn’t wrong. It’s just choosing based on what it thinks is normal.
For personal projects and low-risk static sites, the stakes stay low enough to experiment and learn. But for anything involving user data or a production service, Mueller made it clear: you want someone who understands what they’re doing. That might be you learning the fundamentals, or it might be hiring an expert. Either way, ignorance becomes expensive fast.
How Search Engines Actually See Vibe-Coded Sites
Here’s something that might surprise you: sites built with vibe coding produce reasonable HTML that doesn’t scream “I was AI-generated.” Mueller said: “Practically speaking, nobody can really recognize as being like, this is a vibe coded website.” Search engines aren’t flagging these sites as second-class citizens because of how they were built.
That doesn’t mean there aren’t risks though. Common vibe coding frameworks can leave recognizable patterns. Mueller flagged a specific example: he reviewed a vibe-coded Bento Grid Generator and found crawlability issues, obsolete meta tags, and content stored in JavaScript files that search engines couldn’t access. The code looked fine. The SEO execution wasn’t.
There’s another trap that catches people off guard. Once a site looks polished, it’s tempting to have the AI write the content too. Mueller acknowledged the tool can do that. But he doesn’t see the value there. Splitt agreed. His point: if the content is AI-written, why would someone visit your site instead of talking to the AI directly?
That’s not a technical problem. That’s a business problem. And it matters.
What This Means for Your Next Project
Mueller and Splitt weren’t laying down formal Google policy. They were sharing what they’ve tried and what they’ve run into. The message for people testing these tools is straightforward: AI handles code generation well, especially for lower-risk projects. It doesn’t make SEO decisions on its own. Those still require someone who knows what to ask for.
If you’re building something simple and personal, AI can save you weeks. Go for it. But if you’re building something that needs to be found, needs to convert, or needs to handle real user data, the calculus changes. You need technical direction. You need someone thinking about crawlability, canonicals, robots.txt, and all the boring stuff that actually matters to search engines.
The good news: that someone doesn’t have to be a 10-year veteran. It just has to be someone who knows what to ask the AI for. And now you do.
FAQs
Can AI-built websites rank in Google?
Yes, they can. The code itself doesn’t disqualify a site from ranking. But the site still needs proper SEO fundamentals like crawlable HTML, correct canonicals, and a working robots.txt. AI won’t implement those automatically without explicit direction.
Should I let AI write my website content?
Not as your main strategy. AI can produce readable content, but it raises a real question: why would someone visit your site instead of asking the AI directly? Your content needs to offer something unique your perspective, your experience, your specific expertise that the AI tool itself doesn’t already provide.
Do I need to know how to code to use vibe coding tools?
You don’t need advanced coding skills, but technical understanding genuinely helps. Knowing what site generator you want, how to structure your setup, and what SEO fundamentals matter will produce much better results than just describing what you want in plain English.
What’s the biggest SEO risk with AI-built sites?
Storing important content in JavaScript files that search engines can’t crawl, using outdated meta tags, and building sites without thinking about canonicals or robots.txt rules. These aren’t problems with AI code quality. They’re problems with not giving the AI enough technical direction.
Is it worth hiring a developer instead of using vibe coding?
It depends on your project and budget. For simple, low-risk sites, vibe coding saves money and time. For sites that need to rank, convert, or handle user data, you probably want human expertise involved either to direct the AI or to build it from scratch with SEO in mind from day one.
