Effective date: 18 July 2026 Controller: Prepseen (“we”, “us”) Contact: privacy@prepseen.app · Hakneset Hagdola 20, Tel Aviv, Israel
Prepseen is an interview-practice coach. Its guiding principle: keep the most sensitive data — your voice recordings — on your own device, while syncing your written practice data to your account so it follows you across devices. This policy is specific about what stays local, what we store, and why.
Prepseen offers two inference modes, and your privacy exposure depends on which you use:
Free / “bring-your-own” mode (local CLI): AI questions, company research, and feedback run through the Claude command-line tool on your own machine, under your own Anthropic login. In this mode your prompts do not pass through Prepseen’s servers at all; they go from your machine to Anthropic directly. Prepseen receives nothing about the content.
Paid / “managed” mode (credits): so you don’t need your own Anthropic key, your prompts are sent to Prepseen’s inference proxy, which forwards them to Anthropic and returns the result. This is the only mode in which your prompt content touches our infrastructure — transiently, to produce your feedback. We do not persist it (§4).
You choose the mode; in hosted production builds the managed mode is the default.
Keyed to an account identifier from our authentication provider (see §7):
| Data | Purpose | Notes |
|---|---|---|
| Account identifier (opaque user ID) + email used to sign in | Create your account, sign you in | Email is held by our auth provider (Clerk); we use magic-link sign-in |
| Synced practice data — your roles, sessions (including question transcripts and AI feedback), and prep notes/stories | So your written practice data follows you across devices | Stored on our servers keyed to your account. May contain your transcribed answers and notes about real, named people you research. This is personal data — covered by your export/delete rights (§10) |
| Credit balance & ledger | Track credits you’ve bought and used | Numbers only |
| Usage metadata | Meter inference and prevent abuse | Action type, token counts, credit amounts, timestamps |
| Purchase records | Accounting, tax, refunds, fraud/chargeback handling | Pack, amount, processor transaction ID, timestamps |
| Product-analytics events | Understand the buy funnel and reliability | Metadata only, via PostHog (§7) — never transcript/answer content |
| Operational logs | Security, debugging, abuse prevention | May include IP address and request metadata |
We do not store your audio recordings — those never leave your device (§5). And your written data is stored only to sync it to you; we don’t sell it or use it to build profiles of the third parties you research.
When you run analysis on the managed plan, the prompt we send for inference may contain content you’d consider sensitive — your transcribed answers, and context about a real, named interviewer or company you asked us to research. To be accurate about this:
If you want a fully local experience where nothing reaches our servers, the
self-host build (npx prepseen) runs entirely on your machine with local file storage.
The hosted service at prepseen.app is the managed, synced experience described here.
Prepseen records your spoken answers and transcribes them on your device (local Whisper). The audio never leaves your device — it’s kept in your browser’s local storage on the device where you recorded it, and we never receive or store it. (Because audio is device-local, recordings don’t sync: they’re available only on the device that made them.)
The transcript (the text of your answer), by contrast, is saved as part of your session and synced to your account (§3) so your practice history is available across devices. You can delete on-device recordings, and export or delete your synced sessions, at any time (§10).
A core feature researches the company and, where you ask, the named interviewer(s) so practice questions and feedback can be tailored. Please use this responsibly:
We share the limited data above only with providers that run the service:
us.i.posthog.com).Prepseen uses very few cookies, and no advertising or cross-site tracking of any kind.
| Category | What / who | Purpose | Consent? |
|---|---|---|---|
| Strictly necessary | Authentication cookies set by Clerk (session cookies) | Keep you signed in securely | Not required — essential |
| Strictly necessary | Local storage set by Prepseen (theme, inference mode, and your cookie choice itself) | Remember your preferences on this device | Not required — essential |
| Payment (checkout page only) | Cookies set by Paddle when you open the checkout tab | Process your payment and prevent fraud | Not required — essential to the purchase you request |
| Analytics (optional) | Vercel Web Analytics | Aggregated, cookieless page analytics — no cookies set, no cross-site tracking, no advertising profiles | Loaded only if you choose “Allow analytics” in the cookie banner |
Depending on where you live (GDPR/UK GDPR, CCPA/CPRA, and others) you may have rights to access, correct, delete, port, or restrict your data, and to object to certain processing.
Note on analytics deletion: when you delete your account we also request deletion of your associated events from our analytics provider (PostHog), so the erasure is honored there too.
Prepseen is not directed to children under 16, and we do not knowingly collect their data.
We protect data with measures appropriate to its sensitivity, including transport encryption, scoped API keys, rate limiting, and access controls on the proxy. No system is perfectly secure, but keeping interview content on-device by design removes the most sensitive data from our servers entirely.
We may update this policy; we’ll change the effective date above and, for material changes, provide a more prominent notice.
Prepseen · privacy@prepseen.app · Hakneset Hagdola 20, Tel Aviv, Israel