Primary region Frankfurt
The application, database and files are located in Frankfurt; transcription, text recognition and AI evaluation are processed within the EU.
Privacy & security
Protection for particularly sensitive practice data.
Psymira processes psychotherapeutic content as a processor on behalf of the practice. This overview describes the current technical state and replaces neither the data processing agreement nor the privacy policy nor the practice’s own review. The authoritative texts are the German versions.
Professional control
AI creates drafts. Review and decisions remain with the psychotherapist.
Tenant and permission binding
Every access by the application is authenticated, and every data query is bound to practice and user permissions.
Data flow
From the browser to the documented result.
Patient and treatment data are not used for advertising or profiling.
Transmission
Connections between the browser, the app and cloud services are protected via HTTPS/TLS.
Storage
Practice and patient data are kept in a database, audio files in a separate file store. The primary region is Frankfurt.
Transcription and text recognition
Audio is transcribed via Google Cloud Speech-to-Text within the EU; the audio file and the transcript can be deleted separately. Uploaded documents are read via text recognition with Document AI within the EU.
AI processing
Before Vertex AI calls, configurable rules for identifier filtering/data reduction are applied so that identifying master data are reduced. Free text (e.g. notes, transcripts) may still contain names; control over this lies with the practice.
Roles and responsibility
A clear division of tasks.
The practice remains responsible for the lawfulness of treatment, documentation and recording. Gleevity GmbH processes patient data solely on behalf of the practice and on documented instructions.
The practice as controller
Determines purposes, legal bases, retention, access rights and the permissibility of audio recordings.
Gleevity GmbH as processor
Operates Psymira under the data processing agreement and on the practice’s documented instructions.
Own responsibility
Gleevity GmbH is itself responsible for the website, contract account, billing, contact and the security of its own service operations.
Sub-processors and recipients
For processing patient and treatment data on behalf of the practice, Psymira uses Google Cloud (App Engine, Cloud SQL, Cloud Storage, Cloud Run, Cloud Logging), Google Cloud Speech-to-Text for transcription, Google Cloud Document AI for text recognition on uploaded documents, Google Vertex AI for the AI functions and Firebase Authentication for sign-in and token verification. The complete, binding list of these sub-processors is contained in the data processing agreement (Annex 3). Other recipients for processing under our own responsibility (account/sign-in, payment, clinic finder, website contact) are described in the privacy policy. This page deliberately does not keep a separate register that could differ from them.
Storage and deletion
The practice controls the retention of patient records (guided by § 630f BGB). The following technical periods apply:
- Audio recordings
- Deleted after successful transcription, at the latest after 30 days.
- Transcripts
- Can be deleted separately or together with the associated recording or patient record; controlled by the practice.
- Audit logs (access to patient data)
- 12 months, then deleted.
- AI and transcription jobs
- Completed queue entries are deleted automatically after seven days.
- Application logs
- Rolling 30 days in Google Cloud Logging.
- Database backups and point-in-time recovery
- Daily backups and transaction logs with seven days of retention.
The complete overview of retention periods is contained in the privacy policy and Annex 4 of the data processing agreement; the details there are authoritative.
How to read this page
Three kinds of statements — kept apart.
On a page like this, commitments, explanations and references to legal provisions stand side by side. They are not the same, and what distinguishes them is stated explicitly here.
Product fact
A commitment about what Psymira does. It is set out bindingly in the data processing agreement or the privacy policy, which are authoritative.
Audio files: Deleted after successful transcription, at the latest after 30 days.
Authoritative: Data processing agreement (Annex 4), privacy policy
Technical explanation
A description of how a technology works in general. It does not apply to Psymira alone and is not a commitment about the scope of functions.
A transcription system produces a verbatim record of what was said. Whether a sentence is professionally relevant does not follow from it — that is a professional judgement.
Authoritative: general explanation, not a commitment
Legal framework
A reference to a provision. It names the provision but replaces neither legal advice nor the practice’s own review.
Health data are special categories of personal data (Art. 9 GDPR). Processing on behalf requires a contract under Art. 28 GDPR; § 630f BGB (German Civil Code) sets the framework for retaining treatment records.
Authoritative: Art. 9 and Art. 28 GDPR, § 630f BGB
The references name the provision and do not replace legal advice. Whether processing is permissible in a given practice remains that practice’s own review — this page makes no statement on that.
How the workflows work in detail
This page describes safeguards and responsibilities. What actually happens in a work step — and what the AI explicitly does not do — is described in the knowledge section.
Frequently asked questions on privacy & AI
Where are practice and patient data processed?
The primary product data are processed and stored in Google Cloud in Frankfurt; this also applies to AI processing. Transcription and text recognition run within the EU. Firebase Authentication processes authentication data only; this may involve processing in the USA.
Are data used to train AI models?
Psymira uses Vertex AI. Under the applicable Google Cloud service terms, customer data are not used to train or fine-tune models without prior permission or instruction. Psymira does not use practice or patient data for training its own models.
Who can access patient data?
Access is tied to practice and user permissions. Administrative access to production data is limited to a narrowly restricted group of people, takes place only in exceptional cases and is logged. There is no external support provider with database access.
How are AI results controlled?
AI output consists of editable drafts. Professional review, editing and decisions remain with the psychotherapist. Psymira makes no diagnosis and takes no treatment or emergency decision.
What applies to audio recordings?
Recordings may only be made if the practice has checked their legal permissibility and obtained the required consent of everyone recorded. The audio file and the transcript can be deleted separately. For the recording itself: Deleted after successful transcription, at the latest after 30 days.
How long is access to patient data logged?
Access to patient data is logged. Retention of audit logs: 12 months, then deleted.
What happens to the data at the end of the contract?
Former paying customers receive a closing export: 90 days after access is deactivated, then deletion (effective immediately, final after seven days); handed over only after identity verification. For an account from a trial only, without a paid subscription: 30 days read-only after the end of the trial, then automatic deletion (effective immediately, final after seven days); no closing archive.
Are patient data used for advertising or profiling?
No. Patient and treatment data are not used for advertising or profiling. A separate consent dialogue applies to the website; it concerns website audience measurement only and never content from the application.
Privacy questions about getting started?
You can reach our data protection officer at datenschutz@psymira.de.
Get in touch