Google Search Relations team members John Mueller and Martin Splitt discussed vibe coding websites on a recent episode of Search Off The Record.

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.

Telling AI To ‘Add Some SEO’

Mueller compared the experience of vibe coding to working with a developer who doesn’t specialize in search.

Mueller 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.”

Vague instructions produce vague results, whether the builder is human or AI. Mueller said he got better outcomes by telling the system what he wanted from the start. That included the domain name, canonical setup, sitemap files, and a robots.txt.

He checked whether the pages used reasonable HTML and linked properly. He also set up pre-publish checks to verify that URLs returned content and that JavaScript files weren’t blocked by robots.txt.

What They Built

Mueller has been building 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 recently switched from VS Code with Copilot to command line tools. He named Claude Code and Gemini CLI as what he currently uses.

Splitt tried Google AI Studio to build a client-side tool with JavaScript. He described the output as readable, looking like a standard Next.js application. But he hit a loop where the AI kept using a library he didn’t want.

Splitt said:

“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.”

The Technical Knowledge Question

Both acknowledged the tension in vibe coding’s promise that you don’t need to know how to code.

Mueller noted that technical understanding helps at every stage. Knowing what kind of site generator you want and how to structure pre-publish checks produced better results. Without that background, the AI will make assumptions. It might choose a static site generator, a JavaScript-heavy setup, or a full CMS with a database backend.

Mueller said:

“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.”

For personal projects and low-risk static sites, the stakes are low enough to experiment. But for anything involving user data or a production service, Mueller added that you’d want someone who understands what they’re doing.

Vibe-Coded Sites & Search Visibility

The sites Mueller built produced reasonable HTML that wouldn’t stand out as vibe-coded.

“Practically speaking, nobody can really recognize as being like, this is a vibe coded website,” he said, adding that common vibe coding frameworks can leave recognizable patterns.

He also pointed to a related risk with content. Once a site looks polished, it’s tempting to have the AI write the content too. Mueller acknowledged the tool can do that but said it’s not where he sees the most value.

Splitt agreed. AI-written content raises the question of why someone would visit a site instead of talking to the AI directly.

Mueller has flagged similar gaps in vibe-coded sites before. He reviewed a vibe-coded Bento Grid Generator on Reddit. He identified issues with crawlability, obsolete meta tags, and content stored in JavaScript files that search engines couldn’t access.

Looking Ahead

The podcast didn’t include formal guidance or policy positions on vibe-coded sites. Mueller and Splitt were sharing what they’ve tried and what they’ve run into.

For people testing these tools, the message is that AI can handle parts of 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.

Featured Image: YouTube.com/GoogleSearchCentral, May 2026.