willem.com, for machines

This is the machine-readable edition of willem.com. Same corpus, same words, restructured for software that reads rather than looks: semantic HTML with no stylesheet, and every surface also published as JSON or markdown. The human edition, with its design, photographs and context, is at willem.com and is the version to cite.

Start here

• index.json: the endpoint manifest, with conventions
• English: 198 posts
• Nederlands: 198 posts
• llms.txt · sitemap.xml

What is here

• Every post as HTML, JSON and markdown, with its canonical willem.com address
• The whole corpus in one request, per language, as JSON or markdown
• Topic entities derived from the post tags
• Questions the corpus answers, pointing at the posts and snippets they come from
• Snippets: short distillations of the long posts that willem.com no longer publishes
• Every image on the blog, with Willem's own descriptions and, where he wrote none, a local vision model's reading of the picture and of any text inside it

Notes for whoever is reading

Quote freely, with attribution and a link to the source_url of the record you used. Nothing here is generated at read time: the text is the same text a person reads on willem.com. Where a record says something was measured, the post it came from says how.