The framework
A complete assessment framework, built on the official curriculum.
A default progression for each level, backed by a resource bank linked to the curriculum. The teacher modifies it freely. Supported, not directed.
The central design principle is to structure the teaching pathway around proposed assessment activities.
Assessment is not merely a way to measure learning outcomes; it becomes a key driver of the teaching and learning pathway.
A default progression, freely modified
The platform extends the subject curriculum. It does not replace it.
For a given level, the teacher receives a default progression: teaching sequences organised over time, backed by a resource bank. The bank brings together course materials, exercises and assessment items, all linked to the competencies of the curriculum. It hosts existing institutional resources and grows with teachers’ own contributions.
- Diagnostic
- An optional test to check prerequisites before a teaching sequence.
- Formative
- In class and at home: recurring errors identified, learning tracked, targeted advice.
- Summative
- To conclude each teaching sequence.
The teacher is free to modify the sequences, add their own materials and items, and adapt the schedule. They can use AI assistance to mark students’ written work, and exchange exercises, items and advice with colleagues.
The idea is not to direct the teacher through software, but to offer a predefined framework, consistent with the curriculum, that strengthens their practice.
The student’s view: assessment that leads to action
A personal space that brings all subjects together.
- Schedule
- Timetable, homework, and upcoming assessments when the teacher chooses to announce them.
- Progression
- For each subject: what has been done, the current sequence, what to know before it, and what comes next.
- Results
- Marks for every assessment, the class average for reference, and the corrections, always available once an assessment is closed.
- Competencies
- The level of mastery, competency by competency, and how it evolves over time.
- Advice
- When a typical error appears in their answers: a suggestion phrased without judgement, concrete actions, then the announcement of the reassessment where they can show their progress.
Strict rules
- It shows only what is certain: no AI proposal the teacher has not validated, no result awaiting marking counted as acquired.
- It shows no ranking, and no other student’s results.
- Competencies and errors are rephrased for the student, in encouraging language, without institutional jargon.
Formative assessment changes its nature for the student: it no longer ends with a mark, it leads to an action. In time, the space could open to families, read-only, and help the form tutor prepare the class council.
What it changes for the system
- Equips teachers who struggle to build their teaching
- Makes it easier for a replacement teacher to take over
- Passes on every change to the national curriculum to the progressions
- Lets national assessments connect with teachers’ daily practice
Principles written into the architecture
A tool that handles student data and shapes teaching is legitimate only if its principles cannot be switched off by a setting.
Student data protection
Students are mostly minors. Any AI task touching their work runs on local models, under the control of the school or institution. Names are pseudonymised first. No profiling, no commercial use.
The teacher decides
AI proposes, the teacher decides. Every proposal is submitted for validation; a rejected one is not proposed again. The origin of each piece of content stays visible.
No mark without human validation
Assisted marking proposes a mark and justifies it against the marking criteria. The teacher or marker decides. AI takes no decision that commits a student’s path.
Transparency and explainability
Each proposal comes with its justification: the competency, the error, the curriculum expectation. Code, prompts and models are open. Models are measured on real assessment tasks before adoption.
Interoperability and reversibility
Content relies on modern web standards, with JavaScript and JSON as pivot technologies. A school can export its data and content at any time. No user is locked in.
Accessibility and equity
Adjustments for students with special needs, such as extra time or specific fonts, are supported. The tool must run on modest equipment. Model bias is monitored before automated marking goes live.
Resistance to change, actor by actor
Each actor may resist, for different reasons. Naming them makes it possible to plan a lever for each.
| Actor | Likely concern | Lever |
|---|---|---|
| Teachers | Fear of being directed or monitored | Complete freedom to modify; no named data reported without consent |
| Inspectors | Fear of being bypassed | They validate the reference progressions |
| Pedagogical advisers | Fear of losing their support role | The tool as a support for their guidance |
| Institutions | Cost and risk of scaling up | Pilot experiments, open standards |
| Funders | Doubt about the return on investment | Costs kept down by open source, impact evaluation |
| Education theorists | Risk of standardising practice | Involvement in design and evaluation |
| Parents | Concern about their children’s data | Local AI, pseudonymisation, transparency |
| Students | Feeling assessed even more | Useful feedback, the right to make mistakes in formative assessment |