AI Coding Tools and SEO: Lessons from Vibe Coding Experiments
Specify SEO requirements upfront when using AI coding tools to avoid vague results and ensure proper site structure.
Specify SEO requirements upfront when using AI coding tools to avoid vague results and ensure proper site structure.
Summary
Google Search Relations members John Mueller and Martin Splitt explored how AI coding tools can produce functional sites quickly, but still require human‑like technical direction for SEO.
Mueller found that vague instructions to AI—such as “add some SEO”—yield vague results. By specifying the domain name, canonical setup, sitemap files, and robots.txt from the start, he achieved better outcomes. He also implemented pre‑publish checks to verify URLs returned content and that JavaScript files weren’t blocked by robots.txt.
Splitt used Google AI Studio to build a client‑side tool with JavaScript, ending up in a loop where the AI kept using an unwanted library. Both highlighted that AI will make assumptions about site generators, JS‑heavy setups, or full CMS backends unless guided.
The discussion emphasized that AI‑generated content can raise visibility concerns and that vibe‑coded sites may leave recognizable patterns, affecting crawlability and meta tags.
For production services, the speakers recommend having a developer who understands the technical stack and can instruct the AI accordingly.
The key takeaway: AI coding tools excel at code generation, but SEO decisions still need explicit, technical input from a knowledgeable human.
Key changes
- AI coding tools need explicit SEO instructions (canonical, sitemap, robots.txt)
- Vague instructions lead to vague SEO outcomes
- Claude Code and Gemini CLI are used for command line AI coding
- Pre‑publish checks improve SEO (URL content, robots.txt, JS blocking)
- Vibe‑coded sites may use static or JS‑heavy setups, affecting crawlability
- AI‑generated content can create duplicate or inaccessible content
- AI tools can produce standard Next.js or static sites but may loop on unwanted libraries
- Using AI for content raises visibility concerns