How we use AI
MedCards AI prepares drafts and gives feedback. People decide what enters your study.
AI drafts, you decide
Drafts of cards, occlusion masks, variants and translations arrive as suggestions. Nothing enters your deck without your approval. Official cards pass editorial review and a specialist check before they publish.
Formative grading
On free-recall answers, the AI compares your answer with the card's rubric and explains what is missing. The grade is feedback for study, not an official mark. You can always set your own rating.
Provenance badges
Every card shows where it came from: written by a person, an AI draft that you reviewed, or reviewed by the editorial team. The badges show on the card and in search.
Where the data goes
The app sends to Anthropic, in the United States, only the text of the action you ask for: the card, your message or the document excerpt.
Before it sends, the app removes CPF, CNS, phone numbers, e-mail addresses, CRM numbers and dates of birth.
Anthropic does not use this data to train models. Calls use zero data retention where the feature allows it.
MedCards keeps the technical trace for 30 days, or 180 days if you turn on "Help improve the AI".
Tutor threads stay 90 days. You can erase all of it in Account › Data › "Delete my AI history".
How to turn AI off
In Account › Data › AI history, the "Use AI" button turns every AI feature off at once. Each feature asks for your consent again before its next use. Study, spaced review and decks keep working without AI.
Limits
The AI makes mistakes. It gives no medical advice and does not replace guidelines, teachers or clinical supervision. When the AI cannot process a part, the app says so and shows the card's sources.