A learning platform, built from a curriculum

If you came here from aporia.gr

Then you have already used the thing this page is about. What follows is what it is, how it was built, and how the same can be done with material you already have.

If you would rather skip the story, tell me what you have. It takes about thirty seconds.

The search queries were already written

This is the whole idea, and it was not ours. The Ministry publishes every school book with a numbered table of contents, and a unit is titled in the exact words a parent types at nine in the evening. Nobody had to invent the vocabulary. It was sitting in public, in a government repository, waiting to be made addressable.

So the first piece of work was not a website. It was reading 197 books into a map of 5,600 distinct units, by class and by subject, each one tied back to the official chapter it came from.

What the learning platform has today

More than four hundred unit pages across eighteen school books and nine subjects, from primary school through to the final year of high school, and more added every week. It is live at aporia.gr, free to use and free of advertising.

Every page names its unit in the book's own numbering, links out to the official chapter on the Ministry's own site, and carries practice a child can actually do. No school book text is reproduced anywhere, because the licence does not allow it and because copying somebody else's pages is not a product.

Practice that is generated, not typed

A page with twenty fixed exercises is finished the second time a child opens it. Here each unit carries a generator: rules that produce a fresh, correct set every time, with difficulty that moves, and an answer that is derived rather than stored.

That changes what the build can promise. A separate test re-derives the answers on its own, from scratch, and disagreement is a build failure rather than a support ticket. The stricter rule sits underneath it: a unit with no working practice is never written to disk, so the site cannot grow a thin page even if somebody wants it to.

Adding a unit is a row in a table, not new code. That is the difference between a project that reaches four hundred pages and one that stalls at forty.

Built to be read by machines as well as children

Every unit page declares what it teaches, which class it is for, which book it belongs to and which questions it asks, in the vocabulary search engines and assistants already understand. The crawlers that answer questions for people are welcomed rather than blocked, and each was checked page by page.

This is not decoration. We audited twenty four competing pages on the same units: every one of them was competent at ordinary SEO and not one marked up a single exercise, a single unit code or a single book. When an assistant is asked about a chapter of the syllabus, there is nothing for it to stand on. That gap is the opportunity, and it is the same argument behind MACHINE 404.

Designed for a child who cannot settle

Most educational sites are built for the child who was going to do the work anyway. This one assumes the opposite.

Sets are bounded and small, so the end is always visible. One thing moves at a time and nothing animates unless it was asked for. There is a calm mode that removes every decorative effect, and a large text mode, both one tap away and remembered.

No claim is made about anybody's condition and none ever will be. The design simply removes the noise, which happens to help everyone.

Publishing only what somebody is looking for

Five thousand six hundred units were mapped. Publishing all of them would have been the obvious move and the wrong one.

Demand was measured unit by unit before a single page was written, and only about a fifth of the core syllabus shows any real search intent at all. Those are the pages that exist. The rest stay in the index until they earn a page.

It is a strange thing to tell a client that most of their catalogue should not be published yet. It is also the difference between a site with a few hundred pages that are found and several thousand that quietly drag each other down.

The skills a project like this needs

It looks like one job. It is several, and they rarely sit in the same person.

Data extraction, to read a public repository into a map you can trust. Domain reading, to understand a syllabus well enough to teach from it. Exercise design, so generated questions are fair, varied and solvable. Verification, because an exercise with a wrong answer is worse than no exercise. SEO at scale, so hundreds of pages do not compete with each other. Accessibility and interface design, so a distracted eleven year old can use it. Licensing judgement, so what is published is legal to publish.

Twenty five years of industrial design and six of building software meet here, and that combination is why it went from an idea to a live site with hundreds of pages in days.

Who else this is for

The same work fits anybody whose material is good and whose pages are invisible.

A tutoring school that wants its syllabus to be the reason parents find it. A publisher of educational material sitting on decades of exercises. A training centre whose course notes are PDFs nobody opens twice. A professional body with a certification syllabus. A manufacturer who has to teach people to use the product properly.

If you have the content, the hard part is already done. Tell me what you have and you get a written scope saying honestly what is possible, what is not, and what it would take, before anything is built.

What you get

• Your syllabus, course notes or catalogue read into structured data that you own
• One page per unit, written to be found, with no database to look after
• Practice generated from rules, so a page never runs out of questions
• A build that refuses to publish any page whose practice does not work
• Marked up so a search engine and an assistant both know what the page teaches
• Attention first: small bounded sets, one live thing at a time, a calm mode and large text
• Sources cited and linked, never copied, so the rights stay where they belong
• A written scope before anything is built, and every claim checked before handover