Privacy Policy

ZeroLayer AI (“ZeroLayer”, “we”, “us”) provides a real-time interview copilot for macOS and the browser. This policy covers the ZeroLayer desktop app and the website and web app at zerolayerhq.com.

• We collect your account email and the interview content you choose to run the copilot on.
• Audio is transcribed by a secure cloud service (Deepgram) by default, or fully on-device (Whisper) if you choose.
• Your question, résumé, job description, and screenshots are sent to an AI model (Anthropic Claude) to generate answers, via our authenticated server. We don’t store that content.
• Payments are handled by Stripe — we never see your card.
• No analytics, no tracking pixels, no ads, no selling your data, no training models on your conversations, and no bot in your call.

When you sign up with Google or email, we store your email address, a Firebase account ID, and your sign-in method in Google Firebase / Firestore. Your subscription and entitlement (plan, status, current billing period) are stored on your account record and are written only by our server — never by your device.

During a live session, ZeroLayer captures two separate audio channels: your microphone (you) and your Mac’s system/call audio (the interviewer). On Windows and Linux the interviewer channel uses shared screen audio. The call audio still plays normally through your speakers — nothing is rerouted.

Cloud transcription (default): audio is streamed in real time to Deepgram, a speech-to-text provider, which returns text. We send Deepgram the model-improvement opt-out flag, so your audio and transcripts are not retained by Deepgram or used to train its models. Audio is streamed transiently — it is not recorded or stored.

On-device transcription (option): choose “on-device” (Whisper) and transcription runs entirely on your machine — your audio never leaves your device. Temporary audio slices are written to a scratch folder and deleted immediately after each transcription.

The Deepgram key lives only on our server; your app receives a short-lived (~120-second), restricted key per session, requested with your signed-in account.

To produce an answer we send, to Anthropic’s Claude model (Pro users may select alternative models): the detected question/transcript, the résumé and job description you provided, recent conversation context, and — only if you use the screenshot feature — a screenshot image. This goes through our own authenticated server endpoint; your app never holds the AI provider key.

We do not store your prompts, answers, or screenshots on our servers — they are streamed back to you and not persisted. Screenshots are captured locally, sent for analysis, then deleted. The only thing our server records for a session is a usage counter (minutes/ answers, with no content) to enforce the free allowance.

The desktop app saves a profile file (zerolayer-profile.json) in your app-data folder. It holds your setup (including your résumé and job-description text) and your saved interview sessions — the verbatim questions, answers, and transcript. This file stays on your Mac. It is currently not encrypted at rest, so anyone with access to your user account on that machine could read it; keep your device account secured, and delete the file to remove this history. The web app stores only your setup configuration in your browser’s local storage.

Subscriptions are processed by Stripe. We create a Stripe Checkout session tied to your account; Stripe collects and stores your payment details. We never receive or store your card number — only Stripe references (customer and subscription IDs) and your plan status.

We use a small, fixed set of trusted providers to run the product. We don’t sell or rent your data to anyone.

• No analytics, tracking pixels, advertising SDKs, or third-party trackers — anywhere in the app or site.
• We don’t sell, rent, or share your data for advertising.
• We don’t train AI models on your conversations, and we opt out of our transcription provider’s training program.
• No bot joins your call; the overlay is a local window that stays out of screen-shares and recordings.
• Optional crash diagnostics are off by default and only run if explicitly enabled, with secrets and content scrubbed.

• Audio: not stored — streamed and discarded.
• Prompts, answers, screenshots: not stored on our servers.
• Account & billing records: kept while your account is active.
• Local interview history: kept on your device until you delete it.

You can use on-device transcription to keep audio fully local, and delete your local profile/ history at any time. To request access to, export of, or deletion of your account data, email us at hello@zerolayerhq.com and we’ll act on it within a reasonable time.

All provider API keys are held server-side, and every transcription and answer request is authenticated with your signed-in session. Data is encrypted in transit (HTTPS/WSS), and database access is gated by per-user security rules so you can only ever read your own account data.

ZeroLayer is not directed to children under 16, and we don’t knowingly collect their data.

We’ll post any changes here and update the “last updated” date above. Material changes will be highlighted.

Questions about this policy or your data? Email hello@zerolayerhq.com.