Schools & Training Organisations

Teaching with AI
without compromising on ethics.

ELYSÉA CERVEAU integrates into your learning tools as an ethical pipeline — posture configurable to your school's values, memory under consent, non-deactivatable prohibitions. The AI does not evaluate, does not judge, does not replace the instructor.

Request a quote SDK Documentation

What you integrate

Three immediate effects on your learning setup — inscribed in the SDK, not in a policy you have to maintain.

Configurable pedagogical posture

The AI's posture — its phrasing, refusals, escalations — is configured to your school's values and method. Pedagogical identity stays at the centre. ELYSÉA does not dictate it — it respects it.

Consented pedagogical memory

The learner's learning context — completed modules, retention signals — is available at each session under their explicit consent. The AI does not start from scratch.

AI Act points handled in the pipeline

Art. 50 (transparency), refusal of emotional evaluation, non-substitution of the instructor — designed to be applied by the pipeline at every call that goes through it; result measured during the campaign, published afterwards, favourable or not. The school's compliance assessment remains its own.

What your learners experience

The AI in your application behaves according to ELYSÉA standards — not because you wrote a policy, but because it is in the pipeline.

Support without dependency

No affective prompting, no engagement optimisation. The AI helps to structure, clarify, and progress — it does not create an artificial need to return.

Clarity on what it is

At every relevant interaction, the AI identifies itself as an AI pipeline. It does not impersonate a human instructor. This boundary is in the architecture — not in a FAQ.

Redirection in sensitive zones

Distress, critical situation, medical or psychological question: ELYSÉA redirects to a qualified human. This prohibition is non-deactivatable, even by the partner school.

What the pipeline is designed to prevent

“ELYSÉA does not evaluate, does not judge, does not replace the instructor.”

These prohibitions are designed to be applied by the pipeline at every call that goes through it; they cannot be disabled by the partner school. They will be measured on corpora during the campaign (result published afterwards, favourable or not): no protection is infallible, and known limits are published on the Limits page.

No diagnosis — regardless of how the learner phrases their message
No emotional evaluation — prohibited by the AI Act in educational settings
No substitution of the instructor — systematic redirection to a qualified human in critical zones
No algorithmic retention — no engagement optimisation, no forced return mechanics
No unconsented influence — no oriented bias in pedagogical recommendations
No human pretence — the AI identifies itself at every relevant interaction

Pricing

On quote

Personalised pricing based on cohort size, level of support, and integration scope.

Single per-learner fee

Paid by the partner school — no monthly subscription on the learner side.

Configured pedagogical posture

Aligned with the school's values and method — the instructor remains at the centre.

Instructor included in the setup

Tool training for the teaching team, integration follow-up.

Frequently asked questions

Is this a course platform?: No. It is an SDK ethical pipeline that integrates into your existing training applications. It does not replace your LMS or your pedagogical content — it controls how the AI behaves inside your tool.
Can the AI carry our school's identity?: Yes. The pedagogical posture — phrasing, refusals, escalations — is configurable to the school's values. The underlying ethical behaviour (prohibitions) remains non-configurable.
How do we handle AI Act compliance for our AI tools?: Three points (transparency Art. 50, refusal of emotional evaluation, non-substitution) are applied by the pipeline by default, with no ethical rules to code. They will be measured during the campaign, result published afterwards, favourable or not; their known limits are published; the compliance assessment of your tools remains your responsibility.
What about GDPR and learner memory?: Pedagogical memory is activated under the learner's explicit consent. It is erasable at any time, accessible only to the learner. The school cannot access it directly.

Ready to integrate ELYSÉA into your training setup?

Personalised quote per school and cohort. Response within 48 h.