Regulatory chat (beta)

A regulatory assistant for Loi Jardé, GDPR, ethics committees and ANSM

Ask your regulatory question, get a focused, sourced answer — without leaking real cases to a consumer ChatGPT.

How it works

Tight scope, sober answers

  1. 1

    One question, a French scope

    Ask anything about RIPH qualification, CNIL reference methodologies (MR-001 to MR-006), ethics committee submissions, ANSM declarations, ICH-GCP.

  2. 2

    The AI answers within scope

    The assistant is constrained to French academic clinical research regulation. Outside that scope, it tells you so instead of inventing an answer.

  3. 3

    Ephemeral conversation

    No session is persisted by default: leave the page and the history disappears. Suitable for sensitive questions covered by professional secrecy.

What you get

Scoped assistance, not a generic ChatGPT

Restricted regulatory scope

The AI is instructed to answer only on Loi Jardé, health-data GDPR, ethics committees and ANSM — outside scope, it says so.

French and EU sources

Loi Jardé, French public health code, CNIL reference methodologies, EU Regulation 536/2014, ICH-GCP: the knowledge base is kept current.

Confidentiality by default

No conversation storage, hosting in France, no data sent for model training. Designed for hospital usage.

Beta access for research offices

Access is being opened progressively to partner research office teams. V1 will explicitly source each answer from the RAG knowledge base under construction.

Frequently asked questions

What you should know before using the regulatory chat

Is the regulatory chat available to every user?

Not yet. The beta is open in priority to partner research office teams (project leads, methodologists, DPO). Investigators will get access once the RAG source base is consolidated.

Are my questions stored?

No, by default. The conversation is ephemeral: it does not leave your session. This is intentional — regulatory questions touch live, sometimes confidential projects.

Which questions will the assistant refuse?

Anything outside the scope of French academic clinical research regulation: clinical medicine, non-health law, taxation, prescriptions. The assistant says so explicitly instead of improvising.

How do you avoid hallucinations?

V0 uses a strict system prompt and refuses to answer when uncertainty is too high. V1 will add a RAG sourced on official texts (CNIL, public health code, Loi Jardé, EU Regulation) to cite the source on every answer.

A regulatory question? Ask it in a controlled context

Sign in to your Protocolis workspace and open the regulatory chat from your study — without leaking your case to a consumer AI.