Zapx Labs
A Zapx Labs browser extension, desktop app and Android app · Last updated: 24 October 2026
Mock Interview Copilot ("the extension") is an educational tool that helps you practice for interviews by transcribing a web meeting you are in and suggesting answers grounded in study notes you provide. This policy explains what the extension accesses, what it sends, and where it goes.
storage.local) on your own device.The desktop app (macOS, Windows, Linux) is the same software packaged as a native application, and this policy applies to it in the same way. Two differences: when you click "Start listening" it captures system audio (the sound your computer is playing, which is how it hears the other participants in any meeting app) and, for the optional Vision feature, a screenshot of your primary screen; and it stores your settings, keys, notes, and transcripts in the app's data folder on your computer instead of browser storage. It sends the same anonymous usage counts described below, which you can switch off in the same place.
The Android app (Google Play, com.zapxlabs.mic) is the same software in a phone app. It listens through the phone's microphone, keeps your settings, keys, notes and transcripts on the phone, and sends the anonymous usage counts described below. Unlike the extension and the desktop app, it also uses PostHog (PostHog Inc., data held in the EU) for product analytics, under the same Privacy switch in Setup; turning the switch off stops all of it.
tabCapture API.The extension sends data only to the third-party providers you enable by entering their API key, with one exception you can see and switch off: in practice mode, when no transcription key is set up, your browser's own speech recognition can write down your spoken practice answers (see the table). You control which are used. Nothing you say, type or upload is sent to Zapx Labs; the only requests to our own server are the anonymous counts and, in practice mode, a download of the interviewer list.
| Data | Sent to (only if you configure it) | Purpose |
|---|---|---|
| Meeting/microphone audio | OpenAI, AssemblyAI, Deepgram, or ElevenLabs | Speech-to-text transcription |
| Your spoken answers in practice mode, only while the "Use Chrome's speech recognition" option is on (offered, switched on, when no transcription key is set up; turn it off to type instead) | Your browser's speech recognition service — Google for Chrome, Microsoft for Edge | Turning practice answers into text |
| Detected questions + your knowledgebase excerpts + transcript context | OpenAI, Anthropic (Claude), or Google (Gemini) | Generating answer suggestions |
| Knowledgebase text | OpenAI | Creating embeddings for retrieval |
| Uploaded PDF files | OpenAI (file upload + vector store) | PDF search |
| Meeting screenshot (Vision only) | OpenAI, Anthropic, or Google | Interpreting shared visuals |
Your use of each provider is governed by that provider's own privacy policy and terms: OpenAI, Anthropic, Google, AssemblyAI, Deepgram, ElevenLabs.
Practice interviewers. The practice room downloads the list of interviewers, interview types and difficulty levels, and the interviewers' pictures, from mic.zapxlabs.com, about once an hour. The request carries the app version and nothing about you: no identifier, no settings, no content, and we do not store who downloaded it. The pictures are portraits of fictional people, most of them generated with AI. Interviewers you create yourself, including any photo you give them, are stored only on your device and are never uploaded.
All of the following are stored locally in your browser only, using Chrome's storage.local, and are never transmitted to Zapx Labs: your API keys; your settings and preferences; your uploaded knowledgebase content and its embeddings; recent transcript snippets and saved transcripts; practice sessions and the interviewers you create for them, with any photo you add; the HUD panel position and size.
You can remove this data at any time using the extension's "Clear interview transcript" control, clearing the extension's storage, or removing the extension.
With your permission, the extension may occasionally show a local desktop notification reminding you that the project accepts donations. These notifications are generated on your own device, are shown at most occasionally, can be dismissed or turned off ("Don't show again"), and involve no server and no data collection.
If you sign in on this website for a managed key, Zapx Labs holds your email address, the keys we issued you (stored only as a one-way hash), your plan, your credit balance, and per-request usage records (model, token counts and cost) needed for metering. If you arrived through a partner link, the code from that link is kept with your account too. While you use a managed key, the detected questions and the knowledgebase excerpts the extension selects are sent to our server and forwarded to the model provider (OpenAI, Anthropic or Google via Vercel AI Gateway) to produce transcripts and answers. We do not store the content of those requests or responses. Your notes, transcripts and settings still stay on your device. You can replace your key at any time from your dashboard, which revokes the old one. To delete your account, email labszapx@gmail.com.
Some links to this site, such as one shared by a creator, end in a short code (?ref=name). We keep a daily count of visits per code with nothing about the visitor attached: no cookie, no identifier and no IP address. If you create a managed-plan account after arriving through such a link, the code is saved with that account so we know which partner introduced you. Nothing is stored in your browser for this. We may tell a partner how many visits and sign-ups their link brought, but not who they were.
Pro subscriptions and credit top-ups on the managed plan, and any voluntary donations, are processed by Stripe under Stripe's own privacy policy. Zapx Labs receives a Stripe customer identifier and payment status but never sees or stores card details.
The download page offers to email you when a new version is released. This is entirely optional. Apart from signing in to the managed plan and the ad measurement described below, it is the only thing on this site that collects personal data. It exists because the desktop app has no automatic update mechanism, so email is the only way we can tell you a new build is available.
send.zapxlabs.com.The app reports a small number of counts so we can tell whether it is working and what to build next. It is on by default and there is a switch to turn it off under Privacy in settings; turning it off stops it entirely.
What is sent: a random identifier generated on your device when you install, the app version, which operating system family you are on, which of the three modes you chose (Zapx key, your own keys, or a local model), and events with no content attached — installed, opened, session started, session ended with its length in seconds, an answer suggested, an answer that failed with the kind of failure (for example "key" or "timeout", never the message), notes indexed with a count of them, and uninstalled. In practice mode: the practice room opened or unable to start and why, a practice started with its number of questions and whether answers are spoken or typed, each answer given with its question number, a step that failed and the kind of failure, a practice left before the end and the step it was on, a practice finished with its length and number of answers, and an interviewer of your own saved, with how many you now keep (never its name, description or picture).
What is never sent: your transcript, the questions asked, the answers suggested, your CV, your notes, your screen, your audio, your API keys, your email, your name, or your IP address. The payload has no free-text field, so there is nowhere for any of it to leak through, and anything we do not recognise is discarded rather than stored.
The identifier is random and is not linked to your Zapx account, your email, or your key (except in the Android app when you add a Zapx key, as described above) — we can see that some number of people run local models, not which people. Reinstalling creates a new identifier, so these are counts of installs rather than of individuals.
Where it goes: to our own server, which stores it and passes exactly the same counts on to PostHog (PostHog Inc., data held in the EU), the analytics service we use to chart them. The extension and the desktop app never contact PostHog themselves; our server does, without your IP address, and PostHog builds no profile from it. Turning the switch off stops both.
We advertise Mock Interview Copilot on Google Search, Quora and Reddit, and sometimes inside ChatGPT. To learn whether those ads bring anyone here, this website can load four measurement tags, plus PostHog for site analytics (below): the Google Ads tag from Google LLC, the Quora Pixel from Quora, Inc., the Reddit Pixel from Reddit, Inc., and OpenAI's Measurement Pixel. They are part of the website, not the extension or the desktop app, which never load any of them.
_gcl_au (a random browser identifier, 90 days) and, if you arrived from a Google ad, _gcl_aw (the ad click that brought you here, 90 days); the Quora Pixel sets no cookie on this site, though Quora may read its own cookies on quora.com if you are signed in there; the Reddit Pixel sets _rdt_uuid (a random browser identifier, 90 days); OpenAI's first-party cookies __oppref (an identifier for the ChatGPT ad click, 30 days) and __obref (a random browser reference, 365 days); and your choice in this browser's local storage.ph_…_posthog, 1 year). If you sign in, your dashboard tells it your account's number (never your email) so a visit before and after sign-in can be joined; signing out clears that. Data goes to PostHog Inc., held in the EU, under the PostHog privacy policy. No thanks switches it off too.
From the extension and desktop app in bring-your-own-key mode: none. Zapx Labs operates no server for the software in that mode and receives no user data, transcripts, audio, screenshots, API keys, or usage information from it. On the managed plan we receive what the Managed plan section describes. Otherwise the only personal data we hold is an email address you have deliberately given us on the website and confirmed, as described above.
Recording, transcribing, or capturing a meeting may require the consent of all participants and may be regulated by law in your jurisdiction. You are responsible for obtaining any required consent and using the extension only in settings where such assistance is permitted (for example, mock interviews and practice sessions). The extension displays an in-app reminder to this effect.
The extension is not directed to children and is intended for adult use in educational and professional-practice contexts.
If this policy changes, the "Last updated" date above will be revised and the new version published at this URL.
Questions about this policy: labszapx@gmail.com · Support