The site you are reading right now
CitationLab runs four country domains in four languages from a single codebase. It is the same build we sell as an AI website, and the most honest reference we have: if the method did not work, it would be visible here first.
One codebase, four country domains, four languages.
The challenge
Four markets, four languages, and one small team. The naive version of this is four separate websites that drift apart within a year — different structure, different claims, different quality, and four times the maintenance for the same message.
There is also a specific hazard in being an AI visibility agency: four near-identical Nordic sites competing with each other for the same answers is not a portfolio, it is a self-inflicted duplication problem.
So the domains had to behave as one system with four faces — one source of truth for structure, one canonical address per language, and no page competing with its own translation.
What we built
One system serving four markets, with the AI-facing surfaces treated as first-class parts of the build.
Four domains, one codebase
citationlab.no, .se, .dk and .ai are the same application. The domain decides the language; nothing is maintained twice.
Localisation
Norwegian, Swedish, Danish and English throughout, plus a Chinese section on .ai for Chinese companies going west.
Canonical and hreflang
Every page has one canonical address per language, with hreflang tying the four together and legacy paths collapsing into the current one.
Agent surfaces
llms.txt, agents.md and a machine-readable profile, announced in the response headers, so an agent arriving at the site knows how to read it.
Structured data
Organisation, services, articles, FAQs and breadcrumbs marked up per locale, so machines read the same facts a human reads.
Content engine
The Citation Journal, guides, a glossary and industry pages — the substance a model needs in order to have something to quote.
Forms and measurement
Lead forms and tracking run through CitationLab Manager, so what the site produces lands in the same system we report from.
AI labelling compliance
Every image in the repository is classified against the AI Act's transparency rules, and a test fails the build if a new one slips in unclassified.
Delivered
Four country domains in four languages from one codebase, server-rendered on the edge, with the agent-facing surfaces built in rather than bolted on.
How we built it
The same three principles we apply to every AI website — applied to ourselves first.
Built with AI
The site is built with AI and Claude Code. That is how a four-market presence stays maintainable for a team this size.
Built for AI
Structure, entities, structured data and agent files are part of the architecture, not a plugin added at the end.
Run by AI
Agents ship changes to this site continuously, and CitationLab Monitor tells us what the models actually say about us afterwards.
Why this matters
It is easy to sell AI visibility. It is harder to be the agency whose own four domains hold up when a model reads them — and that is the version a prospective customer can check without asking us anything.
Everything on this site is a working example of the method: how the pages are structured, what is stated in text, how the languages relate to each other, what an agent is handed when it arrives. If we recommend it, we run it.
The best reference we have is the page you are standing on.
Frequently asked questions
Because the four markets are genuinely different — different competitors, different search behaviour, different regulation. Country domains let each market be its own entity while a single codebase keeps structure, quality and speed identical across all four.
No. Each page has one canonical address per language, and hreflang tells the search and AI systems which version belongs to which market. Legacy paths collapse into the current one, so no page ends up competing with its own translation.
Yes. The multi-domain, multi-language setup is the same AI website product, just at a larger scale. Get in touch and we will go through your markets and what it would take.
Want the same build for your business?
One market or four. Tell us where you want to be visible, and we will tell you what it takes for AI to recommend you there.
Was this page helpful?