PsychSupervisor AI

Privacy Policy

Version 1.8 — 3 November 2026

This document explains which data PsychSupervisor AI processes, why, for how long it is kept and who it is shared with. It describes what the product actually does; it is not a generic template.

1. What this product is and who uses it

PsychSupervisor AI is a working tool for mental health professionals: it transcribes their own sessions and prepares a structured clinical report based on the chosen therapeutic approach. The account holder is the professional. Clients are not users of this product; client data enters the system only when the professional uploads it.

The tool provides clinical decision support. It does not diagnose and does not replace professional judgement.

2. Roles: who is responsible for what

For client data, the controller is the professional who uploads it. We process that data on the professional's instructions and only for the function of the tool. Obtaining the client's consent and complying with applicable law is the professional's responsibility.

For the professional's own account data (name, email, phone) we are the controller.

3. Data we process

Account data

Clinical data uploaded by the professional

Usage and security data

Local assistant (only if you use it)

Data we do not process

4. Why we process it and the legal basis

No data is processed for advertising, profiling or resale.

5. Audio recordings and retention

A raw recording is kept for two purposes only: producing the transcript and letting you review it.

6. Where data is processed

Hosting, database, file storage and the server run on Google Cloud and Firebase in the European Union region (Belgium, europe-west1). Converting audio into text also runs on our own infrastructure in the same region; the recording is not sent to a third party for that purpose.

7. Who we share it with

Data is never sold. It is shared only with the service providers needed to run the tool, and only to the extent needed:

Data may be disclosed where there is a legally binding request from a competent authority. In that case you are informed unless we are legally prohibited from doing so.

7a. Website analytics and cookies

On the website and the web version of the app (psychsupervisor.com), Google Analytics is used to see, in aggregate, how the site is used. It runs only if you consent: before you do, no request is sent to Google and no analytics cookie is set. The Android app has no such analytics.

You can withdraw your consent at any time: choose "Decline" under the "Cookie preferences" link at the bottom of the privacy page (psychsupervisor.com/privacy). Analytics stops and the analytics cookies are deleted from your browser.

8. Artificial intelligence and training

Nothing you upload is used to train artificial intelligence models. We require the providers we use not to use data sent through their programming interface for training purposes.

A report is the output of a model and can be wrong. The product ties every heading of the report to its basis in the transcript; checking that tie is the professional's work.

The local assistant's model (section 8a) runs on your phone and is not trained with your questions or reports.

8a. Optional local assistant

Under Settings > On-device Assistant you can download an artificial intelligence model that runs on your phone (Gemma 4 E2B by Google, Apache-2.0 licence). If you do not download it, the app works as before. The assistant does not run on our servers; in your account it carries out only the actions you approve (see below).

9. Masking and review

Personal data in the transcript is masked automatically. Masking works in two ways:

The mapping between the real name and the name put in its place is not recorded anywhere; it is gone once the job finishes. Masking cannot be undone.

Masking is not perfect: an expression can slip through. That is why the transcript is presented for the professional's review, and can be corrected, before it is sent for analysis.

**The AI chat and the case file are OUTSIDE this path.** There the text is not produced by a machine but written by the professional: the question you type, the case information you give and the session report you add go to the analysis provider unmasked. The reason is the integrity of the text: changing a name inside the professional's own sentence would change what they wrote without their knowledge. On this path the professional knows what was written, and whether to type a real name is their decision; the app states this permanently under the chat box.

The chat and the documents added to a case file are stored under the client file and are deleted together with it.

10. Security

No system is absolutely secure. If an incident affecting data security occurs, notification is made within the period required by law.

11. Your rights

You have the right to access your data, to ask for its correction or deletion, to object to processing, to receive your data in a portable form, and to withdraw a consent you have given.

Most of this you can do yourself inside the app:

For written requests: info@probeside.com

12. Children

The product is not intended for use by anyone under eighteen. A professional uploading client data is responsible for obtaining the required consent from the legal representative if the client is a minor.

13. Changes to this document

When the text changes, the version number increases. For a significant change the app asks for consent again, and the version you accepted is recorded.

14. Contact

info@probeside.com

You can also use Settings > Feedback inside the app.