The real questions about the SDK and its commitments.
Direct answers, unfiltered.

Does ELYSÉA replace my AI model?

No. ELYSÉA CERVEAU does not provide an LLM model: the pipeline sits between your application and the model. Default model: Claude (Anthropic). The brain pipeline requires the builder’s own model key (BYOK); key deposit is not open yet, so the pipeline is not yet open to builders.

How long does integration take?

sk_test_ issued when the developer portal opens — no credit card, capped at 500 interactions per month. The production configuration (sk_live_) goes through an individual review of your use case by the founder of ELYSÉA. You leave with a versioned, auditable canons file.

Does the AI Act apply to me?

Probably, if your app interacts via LLM with people in Europe — the precise assessment remains yours. The ELYSÉA pipeline applies three AI Act-related points by default: AI identification (Art. 50), refusal of manipulation (Art. 5), refusal of emotional evaluation in educational settings. It does not make your product compliant on its own.

Will my users see ELYSÉA in my interface?

Yes. The "Powered by ELYSÉA" label is present in all integrations — a non-negotiable decision. The visual presentation can be adapted to your identity on paid tiers, but the ELYSÉA link always remains visible. The HTTP technical marking is active in all cases — your users can always verify the pipeline status.

Can I add my own ethical rules on top of the 8 prohibitions?

Yes — this is the tightens-only principle. You can configure stricter rules for your domain: linguistic posture, domain refusals, human escalations. What you cannot do: weaken the 8 base prohibitions. Ethical configuration is unilaterally virtuous — upward only.

Can ELYSÉA weaken my product or limit my use cases?

No — ELYSÉA intervenes on the 8 prohibitions. Your business logic, UX and local canons remain yours. What you cannot disable: the ethical canons.

How many messages are included?

Each plan includes 200 messages per active user per month; beyond that, €0.005 excl. VAT per message, billed pro rata. Sales opening soon. No payment is accepted yet.

Can I cancel?

Yes, at any time, without commitment or penalty. No lock-in on your data or configuration.

What exactly is ELYSÉA CERVEAU?

It is a layer of rules placed between an application and its AI — 8 prohibitions applied by the pipeline at every exchange. Builder configuration cannot disable them. Their effectiveness will be measured on corpora during the campaign and published afterwards, favourable or not; known limits are published: no protection is infallible.

Does this site collect my data?

Without consent, no analytics cookie or tracker is set. After explicit consent, PostHog (usage analytics) is enabled — only outside exchange areas. When enabled, a server error log (Sentry) runs server-side, independently of consent, with no personal data by default. The contact form sends your messages via Brevo (hosting region not proven) and is used only to reply to you.

What does "Powered by ELYSÉA" in an application mean?

8 prohibitions are applied at every exchange and cannot be disabled by the builder. The pipeline is designed to refuse diagnosis, manipulation, creating emotional dependency and passing itself off as human. These protections will be measured during the campaign, result published afterwards, favourable or not; their known limits are published on the Limits page.

Does my memory in an ELYSÉA app truly belong to me?

Yes — viewing, JSON export and immediate deletion are available in your space. Prohibition P2 blocks a builder's access to your memory without your explicit consent recorded in the registry.

Do different ELYSÉA apps share my data?

Your memory is cross-app, but it belongs to you. It travels with you in the ELYSÉA ecosystem only with your explicit consent — not activated by default. Every app access is recorded in the registry. You see who accessed what, and you can revoke at any time.

What is the AI Act?

The European AI regulation, being progressively applied since 2024. It requires transparency, restrictions in sensitive contexts and prohibition of certain forms of manipulation. ELYSÉA applies protections in its pipeline that go in this direction; on their own they do not amount to compliance.

Can I trust the label if I don't know ELYSÉA?

The label is tied to the application's SDK access status: to display it, the app must go through the ELYSÉA pipeline. If access is revoked, the label disappears from the interface. What the pipeline applies, and its measured limits, are published on this site.