← SundaMind

Security and data

A diary is worth only as much as the trust you place in it. This page sets out where the data sits, who processes it, and how an account can be removed for good. The privacy policy is the authoritative version.

Where the data sits

Where AI is involved

Three features use a language model: the Companion, the pattern review and the experiment suggestions. The relevant input is sent per request to the interface of Anthropic PBC in the United States. It is a single provider, not a chain.

The Companion is deliberately kept narrow: it reflects and asks. It offers no interpretations and certainly no diagnoses — that boundary is written into the prompt and watched by tests.

If a text suggests a crisis

When the app detects signs of an acute crisis, it immediately shows emergency resources. Detection is deliberately set wide: a false alarm is uncomfortable, a missed signal would be worse.

That prompt is a pointer to help — not an assessment, and not a diagnosis.

Deleting your account and data

You can delete the account yourself in the settings. This removes the associated data in one pass — entries, check-ins and reviews — not merely your access. You confirm by typing your own email address.

Feedback submitted under an explicit promise of anonymity is anonymised rather than deleted: the personal reference is removed while the entry itself remains.

Individual entries can be deleted at any time inside the app. The rights to access, rectification, restriction, portability and objection apply independently; the privacy policy sets out how to exercise them.

What is SundaMind? · About Dirk Werner · Frequently asked questions · Deutsch

SundaMind is no substitute for psychotherapy and is not a medical device. In an acute crisis call 112 (across the EU) or your local emergency number. In Germany, Telefonseelsorge offers free support in German on 0800 111 0 111.