Does the AI make the decision for us?

No. The assistant suggests — leadership accepts manually. That's a deliberate design choice from OECD TALIS AI-governance recommendations and ensures co-determination with the union rep is not compromised.

What data does the AI see?

Only the allocation's own data: teachers, classes, subjects, competence, wishes, budget. Pupil data is not sent to the AI. We use OPENAI gpt-4o-mini with strict JSON output.

Can the AI propose to take hours from a teacher?

Yes, if you ask it to reduce against a budget cap. But reductions always come with reasoning, and the assistant respects locked assignments and minimum-hour rules.

Does the assistant work in English?

Yes. Chat input and proposals follow the school's locale — Danish on skoleelev.ai, English on mystudent.ai. Both use the same model.

Subject allocation — AI

AI assistant for subject allocation

A conversational assistant that proposes assignments, rebalances and reductions — with teacher wishes, budget and competence as context. The assistant never writes directly: every proposal is accepted manually by leadership, in line with OECD TALIS (2019) AI-governance principles.

Contact

Chat interface

Ask in plain English: 'who can take 7B maths?' — get concrete proposals with reasoning.

Three proposal types

Assign (new), rebalance (move between teachers), reduce (cut to hit budget).

Teacher wishes count

Proposals favour teachers who wished for the subject or class — without ignoring competence.

Budget-aware

If you're over cap, it suggests where to cut — not just where coverage is missing.

Competence respect

Specialism and teaching competence are weighted — it won't recommend a teacher without it.

Suggests, doesn't auto-apply

Every proposal is accepted manually by leadership — the assistant never writes directly.

In the classroom

Last 5 gaps

When 95% is allocated and the rest is hard, the AI points to the most likely fits.

Sick leave between rounds

A teacher drops out — ask for rebalances that hurt the fewest others.

Finance asks for cuts

Cut 40 hours — the AI proposes the least painful places.

For school leadership and for the union rep

For school leadership

The assistant removes the most cognitively heavy work — combining subject, class, competence, wish and budget in your head. Proposals come with reasoning, so leadership can explain the decision in negotiation.

Reasoning per proposal — not a black-box recommendation
Budget-aware: suggests where to cut without breaking coverage
Runs on OPENAI gpt-4o-mini with strict JSON output

For the union rep

The union rep sees the same proposals as leadership and can verify that wishes and competence really weighed in. The assistant never writes directly to the allocation, so negotiation rights remain intact — in line with the DLF union handbook (2022).

Proposals are advisory — not decided assignments
Wishes are documented in the proposal's reasoning
Audit trail for every accepted/rejected proposal

Works with

Allocation suite Validation Budget

Research base

The AI assistant follows human-in-the-loop principles from OECD TALIS and Danish co-determination practice.

OECD (2019). TALIS 2018 Results (Volume I): Teachers and School Leaders as Lifelong Learners. OECD Publishing source
OECD (2020). TALIS 2018 Results (Volume II): Teachers and School Leaders as Valued Professionals. OECD Publishing source
Danmarks Evalueringsinstitut (EVA) (2018). Skoleledelsens arbejde med fagfordeling og kompetencedækning. EVA source
Danmarks Lærerforening (2022). TR-håndbogen: Medbestemmelse, forhandling og lokalaftaler om arbejdstid. dlf.org
Darling-Hammond, L. (2000). Teacher quality and student achievement: A review of state policy evidence. Education Policy Analysis Archives, 8(1) source
Datatilsynet (2023). Vejledning om behandling af personoplysninger om ansatte — løn, kompetencer og vurdering. datatilsynet.dk source

Frequently asked questions

Ready to try it?

Get a free demo for your school — we'll show you the whole platform in 30 minutes.