The Jacob House: A Custom Multilingual Hospitality Website Built for SEO and LLM Retrieval

The Jacob House is stronger than a generic brochure site. It is a custom-generated multilingual hospitality website for a holiday home in Middelburg, built for crawlability, strong editorial presentation, and clear AI-readable property data. The project covers three languages, seven core pages per language, a property gallery, FAQ, booking flow, maps, and machine-readable content layers designed to help both search engines and LLM retrieval systems understand the property correctly.

The stack is deliberately simple. A single Node.js generator writes the static HTML, Markdown mirrors, localized llms.txt files, robots.txt, and sitemap.xml, while the frontend stays plain HTML, CSS, and vanilla JavaScript. That keeps runtime complexity low, makes deployment straightforward, and preserves full control over what crawlers and models actually read.

One Generator, One Source Of Truth

The main architectural choice is centralization. Content, localization, metadata, schema, and page generation all live in one generator, which means one source of truth for page copy, slugs, metadata, language variants, image captions, and structured data. That reduces drift across the Dutch, French, and English versions and makes the output predictable for both search engines and AI systems.

This is what makes the site feel disciplined rather than improvised. The same system controls canonical page structure, shared property facts, localization, and metadata, so the visitor experience and the machine-facing layer stay aligned instead of drifting over time.

Explicit SEO And LLM Optimization

The clearest technical signal is that the site includes explicit retrieval layers, not just traditional SEO. In addition to static HTML output, it ships localized llms.txt files that define canonical identity, verified facts, likely search intents, page entrypoints, content boundaries, and response guidance for language models. That is deliberate LLM-oriented optimization, not accidental side effect.

The generator also writes Markdown mirrors for the site pages, which is useful because many LLM pipelines extract cleaner signals from Markdown than from full HTML with navigation and presentation markup. Alongside that, the project includes structured data, canonical tags, hreflang alternates, sitemap.xml, and robots.txt, so the traditional search layer is just as intentional as the AI-readable one.

Why The Stack Choice Matters

This is not a CMS site and not a framework app. It is closer to a purpose-built publishing system for a single property. That gives the project very low runtime complexity, high crawlability, clean multilingual output, strong control over copy consistency, no dependency on client-rendered content for indexing, and explicit AI-readable surfaces.

The client-side JavaScript stays intentionally light. It handles interface behavior like language selection, galleries, lightbox interactions, reveal animations, and locale-aware booking links, while the actual page content remains readable without JavaScript. In one sentence, this project is a custom Node-generated static multilingual hospitality website with structured SEO, localized AI-readable llms.txt files, Markdown content mirrors, and schema-rich pages designed for both search engines and LLM retrieval.