SEO essentials such as canonicals, sitemaps, and robots. txt files do not automatically configure themselves just because an AI tool generates your website. This key point was emphasized by Google Search Relations team members John Mueller and Martin Splitt during a recent episode of their podcast, Search Off The Record. Both Mueller and Splitt experimented with AI coding tools for their personal projects and encountered a common shortfall. Although these tools quickly produced functional websites, achieving proper SEO still demanded explicit technical input and direction. **Instructing AI to "Add SEO"** Mueller likened working with AI code generation to collaborating with a developer unfamiliar with search optimization. He explained 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 or imprecise instructions lead to equally vague results, whether the coder is human or AI. Mueller found more effective outcomes when he clearly specified requirements from the outset—defining the domain, canonical setup, sitemap files, and robots. txt configuration. He also verified that pages used proper HTML and linked correctly, and implemented pre-publish checks to ensure URLs served content and JavaScript files were not blocked by robots. txt. **Their Projects** Mueller built test websites to observe how Googlebot handled requests, deploying them to Firebase hosting using Hugo as a static site generator with version control via GitHub. Recently, he shifted from using VS Code with Copilot to command-line tools, naming Claude Code and Gemini CLI as current favorites. Meanwhile, Splitt experimented with Google AI Studio to develop a client-side JavaScript tool. While the output was readable and resembled typical Next. js applications, he struggled with the AI repeatedly incorporating an unwanted library. He remarked: “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. ” **Technical Expertise Remains Crucial** They both recognized the tension between vibe coding’s promise of removing the need to understand coding and the practical necessity of technical knowledge. Mueller emphasized that expertise enhances results at every phase—knowing the preferred site generator type and how to establish pre-publish validations improves outcomes. Without guidance, the AI will make assumptions about the architecture, whether a static site generator, a JavaScript-heavy framework, or a CMS with a database backend. Mueller noted: “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 or low-risk static sites, this exploratory approach may suffice. However, for applications involving user data or live production services, Mueller recommends involving knowledgeable professionals. **SEO and Visibility of Vibe-Coded Sites** The sites Mueller created rendered reasonable HTML, making it difficult to discern them as vibe-coded websites. He commented: “Practically speaking, nobody can really recognize as being like, this is a vibe coded website, ” though he cautioned that some common vibe coding frameworks may leave identifiable patterns. He also raised concerns about content: polished sites might tempt owners to rely on AI to generate their content. Mueller acknowledged this capability but felt AI-written content offers limited value. Splitt concurred, suggesting AI-generated content prompts the question of why a user would visit a site instead of querying the AI directly. Previously, Mueller has highlighted similar issues with vibe-coded sites, such as a Bento Grid Generator he reviewed on Reddit, pointing out problems with crawlability, outdated meta tags, and content trapped in JavaScript inaccessible to search engines. **Looking Forward** While the podcast did not offer formal guidelines or policy statements on vibe-coded websites, Mueller and Splitt shared insights based on their experiences. The takeaway for those experimenting with AI coding tools is that AI can efficiently assist certain coding tasks—particularly for low-stakes projects—but SEO decisions still necessitate knowledgeable human involvement. --- *Featured Image: YouTube. com/GoogleSearchCentral, May 2026. *