Your study history is yours.

What the private beta keeps

• Account identity, profile preferences, exam date, and study availability.
• Plans, answers, confidence ratings, review history, lab reflections, notes, and the immutable evidence needed to explain a readiness estimate.
• Pseudonymous rate-limit and AI-budget records used to prevent abuse and cap spend.

AI tutor data

A tutor question and the minimum grounded study context are sent to the configured AI provider to generate an answer. CloudKata does not store the tutor conversation as a learning record. Do not put credentials, employer-confidential data, or personal information in a tutor question.

Providers and operations

Supabase provides authentication and the database, Vercel runs the web application, and the configured AI provider answers admitted tutor requests. The private beta uses privacy-minimized platform logs for operations; request bodies, tutor text, notes, cookies, raw user ids, and email addresses are excluded from application logs.

This first private-beta deployment has no provider-managed database backup. That avoids implying a recovery guarantee the free hosted plan does not provide, but it also means a provider data loss may be unrecoverable. Managed backups and restore testing are required before external beta enrollment opens.

Retention, export, and deletion

Live learner data remains while the account is active. You can export the registered learner data from Profile. You can also request deletion there; the current recovery window is seven days, after which the Auth identity and registered learner-owned rows are permanently removed by scheduled maintenance. Opaque abuse-control buckets expire after seven idle days. Platform logs use the shortest useful provider window and never more than 30 days.

Questions or corrections

During the invite-only beta, use the same private channel through which you received your invitation. Do not send passwords, API keys, or sensitive study/employer material.