The Timetable Was Never the Hard Part

Every AI timetabling product demonstrates itself the same way. Constraints go in, a conflict-free grid comes out, and the room makes the appropriate noise. It is a good demo, and it is answering a question that most schools never get to ask.

Because the timetable is not where schools get stuck. They get stuck three weeks earlier, on a form.

We learned this from a lecturer setting up a research master’s programme — a genuinely complex one, with lectures, tutorials and lab sessions running on different weekly rhythms. She expected to give us her programme documents and get most of the way there. Instead she was asked to enter every subject, its weekly hours and its session types by hand, one field at a time. She abandoned setup before she ever reached the generator.

The generator would have worked. She never saw it. That is the whole problem in one sentence, and it is not a problem you can fix by making the solver cleverer.

Consider what timetable setup actually asks a school for. Every subject in every programme. Weekly hours for each one, which the school usually holds as a semester total that someone has to divide. Session types, because a two-hour lecture and a two-hour lab are not interchangeable. Which groups take what. Teacher assignments and availability. Rooms and their capacities. For a mid-sized programme that is several hundred discrete entries, and the person doing it is a teacher in August.

Now consider that the school already has all of it. It is sitting in the study plan, the module descriptions, the programme handbook — the documents that were written, reviewed and approved long before anyone opened a scheduling tool. The information exists. It is simply in a Word file rather than a database, and every timetabling product on the market asks a human to be the bridge.

So we removed the human from that step. You hand Skoolia the documents you already have, and it does the transcription.

The flow is deliberately unglamorous. You upload your study plans and module sheets. The AI reads them and extracts the structure — subjects, hours, session types, groups. It assembles a draft and shows it to you. Where the documents genuinely did not say something, it asks you a specific question rather than inventing an answer. You review the draft, correct anything wrong, and confirm. Only then does anything get written, and what gets written is exactly the same records the manual wizard would have created. From there the existing generator runs, unchanged.

The unglamorous parts are where the work actually is. Academic documents write hours as 20h, as 1h30, as 20 heures, as a decimal with a comma. French programmes label sessions CM, TD and TP, which have to map onto lectures, tutorials and labs. Semester totals need dividing into weekly loads. The same subject appears in three files with three slightly different names and has to be recognised as one subject. None of that is impressive in a demo. All of it is the difference between a tool that reads your documents and a tool that claims to.

Two design decisions matter more than the extraction itself.

The first is that it asks. A model that fills every gap with a confident guess produces a complete, plausible, wrong timetable — and a wrong timetable that looks finished is worse than an empty one, because nobody checks it until a class turns up to a room that is already occupied. Documents are incomplete by nature; they were written for humans who could infer the rest. So where the source is silent, Skoolia raises a targeted question instead of filling the space.

The second is that nothing is created until you say so. The draft is a proposal. You see every subject and every hour before a single record exists in your school, and you can change any of it. This is what makes handing over your documents a low-risk act rather than an act of faith.

This is why we think Skoolia is the number one AI timetabling platform available today. Not because our solver is a secret — timetable generation is a well-understood problem and anyone serious can do it. Skoolia is number one because we are solving the step that actually decides whether a school ever gets a generated timetable at all. Everyone else optimises the engine. We removed the reason schools never reach it.

A school that can go from its own study plan to a working timetable in an afternoon is in a different category from one that can generate beautifully once someone has spent three weeks on data entry. The second school, in practice, does not generate at all.

Two practical notes. The guided Setup Wizard is now included on every Skoolia plan, not held back for larger schools — the schools with the least admin capacity need it most. And importing from documents is available from Scheduling whenever you want it, not only during first-time setup, because next semester’s study plans arrive every semester.

You can see the full AI timetable generator, the wider AI scheduling capabilities, or the standalone AI Timetabling plan if scheduling is all you need for now.

Bring the study plan you already wrote. That was always supposed to be enough.