
AI content at scale for the tourism of Zeeland
Tourism in Zeeland is multilingual: German, English, French, Flemish, and increasingly Polish, Czech and Italian. Hello Zeeland wanted to scale the editorial team without losing the voice of the region. Together we built an AI-driven content pipeline that delivers drafts in 12 languages, with human approval before every publication.
A handful of editors, a region that wants to be found in six languages at once
Hello Zeeland is the regional tourism organisation behind the online visibility of Walcheren, Schouwen-Duiveland, the Bevelanden and Zeelandic Flanders. The editors write about attractions, events, cycling routes and local businesses, and publish in a growing stack of languages. The question is always the same: how many stories can we ship per month at quality, in how many languages, without it becoming artificial or generic?
Translating the old way takes time and money that does not scale with seasonal peaks. A German pre-season already needs four new routes and ten events on March 1, and at that moment the Czech, Polish and Italian versions are often still idle. The editorial team was caught between quality and speed, and wanted a workflow in which AI does the heavy lifting and the editor guards the voice.
Three steps, one workflow the editorial team can run
Analysis of the current editorial flow
We mapped the existing content flow from brief to publication, per language. Which steps take the most time, where quality leaks, which decisions belong with the editor and which can shift to AI.
Strategy for 12 languages
We set a prioritisation: NL and DE as core languages, EN and FR as the base for international guests, and six emerging markets as a growth track. Per language we defined tone of voice, length and SEO guidelines, in a prompt stack that AI can apply consistently.
Implementation in the editorial team
We built a dashboard where editors enter one brief and receive a draft page in 12 languages. Each language has its own quality checks, source links and an approval button. The editorial team stays accountable and steers on tone, not typing.
An editorial pipeline where AI does the lifting and the editor publishes
No loose tools, but one workflow the editorial team runs every day. Components are loosely coupled so we can adjust per language or per content type without rebuilding the whole.
- 01Editorial workflow with AI drafts and human final editing
- 02Translation pipeline to 12 languages with on-brand tone of voice
- 03Content model for attractions, events and routes
- 04Automated SEO checks per page and per language
- 05Review dashboard where editors approve draft pages
- 06Measurable lead times from brief to publication
What the pipeline delivers in numbers
languages supported from a single brief
faster lead time from brief to publication
pages equipped with consistent multilingual content
* Numbers are indicative and based on the first measurement period
“The editorial team is faster and at the same time less rushed. We have time again for the stories only Zeelanders can write, and AI clears the multilingual chores.”
next case