Privacy Policy

StageMirror records you speaking so it can tell you how you came across. Those recordings are private to your account, they are sent to OpenAI to be transcribed and analyzed, and they are deleted on a schedule you choose. We do not train models on your data, and we do not sell it.

• Your account — email address and, if you set one, a display name.
• What you are preparing for — after you enrol we ask one optional question, and store the option you pick. If you pick “Something else” we also store the few words you type. We use it to decide what to build next, nothing else — it never changes the fourteen days, and you can skip it or change it later in your settings.
• Your recordings — the video and audio of each practice take, plus its duration and the prompt you were speaking to.
• What we derive from them — a transcript with word-level timings, speaking metrics (filler rate, pace, pauses, pitch variation) and camera-based estimates (eye contact, posture, gestures), plus written coaching.
• Billing — if you subscribe, Stripe processes your payment and we store only the resulting customer and subscription identifiers. We never see or store your card details.

OpenAI. To analyze a take we send its audio to OpenAI’s transcription API (the whisper-1 model) and send the resulting transcript text to OpenAI’s chat API to generate coaching. OpenAI processes this as our service provider under their API terms, which state that data submitted through the API is not used to train their models. Video frames are analyzed on our own servers and are never sent to OpenAI.

Stripe processes payments. Resend delivers account emails such as sign-in confirmations. Our database, storage and analysis worker run on infrastructure we operate ourselves.

Plausible counts visits to the website. It sets no cookies, stores no identifier for you on your device, and builds no profile — it cannot follow you to any other site, which is the reason we use it instead of the usual alternative. It records the page you opened, the referrer, and your country, device type and browser as your browser reports them. It also records a handful of named product moments — that an account was created, that a take was analyzed, that a subscription started — and never who did them: your email, your user id and the identifier of any individual take or recording are never sent, and the page addresses that contain one are stripped before the request leaves your browser. Nothing about a recording, a transcript or your coaching ever reaches it. Plausible runs on the website only.

PostHog measures how the product is actually used, on the website and in the mobile app. It is a different kind of tool from Plausible and the difference matters, so we are specific: it stores an identifier for you on your device, and from the moment you sign in that identifier is linked to your account id, so the moments below can be counted per account instead of only in total. Your email address and your name are never sent to it.

It receives the same short list of named product moments Plausible gets, plus the screens you open. It does not watch you use the app: session replay is switched off, and so is the automatic click capture that would otherwise send the text of whatever you tapped — which on this product would mean prompts, day titles and lines of your own transcript. Page addresses that contain the identifier of an individual take are stripped before the request leaves your device. Nothing about a recording, a transcript or your coaching ever reaches it.

RevenueCat handles subscriptions bought inside the mobile app. When you sign in, it receives your account id so a purchase can be attached to the right account, and it receives the purchase record the store gives it — which plan, when it started, whether it renewed. It never receives your email, your name, or anything about a recording. The payment itself is taken by Apple or Google Play, depending on the device: we never see your card, and the store’s own privacy policy governs what it collects at the moment you pay. A subscription bought on the website goes through Stripe instead, and the same is true there.

Google AdMob supplies a non-personalized banner to free users of the mobile app. It may process the device IP address (including to estimate general location), device or app identifiers, ads shown and interactions with them, and SDK diagnostics and performance data for ad delivery, measurement and fraud prevention. The SDK is not initialized until Google’s consent system says an ad request is permitted. Pro users do not receive ad requests.

That is the complete list. We do not use your StageMirror account, recordings, transcripts or coaching to personalize ads, and we do not track you across other apps or websites. Neither the app nor the website contains analytics that follow you off it.

The controller of your personal data is deepfai Inc., which publishes StageMirror. For anything in this policy — including every request described under Your choices and your rights below — write to privacy@stagemirror.ai.

Our servers are in the United States, and so are the providers named above. If you use StageMirror from the UK, the EEA or Switzerland your data is transferred there. Those transfers rely on the European Commission’s Standard Contractual Clauses, which are part of the data processing agreement each of those providers offers and which we have accepted with each of them.

If you are in the UK or the EEA, the GDPR requires us to name a legal basis for each thing we do with your data. Ours are:

• To provide what you asked for (performance of a contract) — your account, your recordings, the transcription and analysis of them, your scores and progress, and the billing that goes with a paid plan. This is the bulk of it. If you do not give us a recording there is nothing to analyze, so this data is genuinely necessary rather than merely useful.
• Our legitimate interests — product analytics, keeping the service up, and detecting abuse. Our interest is understanding which parts of the product work; we weigh it against your privacy by keeping analytics to how the app is used, never to the content of what you say.
• Your consent — the ads shown to free accounts in the mobile app, which is what the consent form on first launch is asking about. You can change that answer at any time in the app, and withdrawing it does not affect anything done before you withdrew it.
• A legal obligation — the payment and tax records we have to keep after a purchase.

Each account has a retention window, set in your settings, between 1 and 365 days and defaulting to 30 days. A scheduled sweep permanently deletes the video and audio of any take older than that window.

On the free plan two things narrow that. Recordings stay playable for 7 days; after that the file is still stored but nothing can open it, including us from the app, until the account moves to Pro. And whatever window you set, free recordings are deleted after 90 days at the outside. The cap can only ever shorten the life of a recording — if you set 14 days, 14 days is what happens, on either plan.

To be precise about what the sweep does and does not remove: it deletes the media files. The session record and the scores derived from it are kept, so your progress over time survives the footage — including through the 7-day lock, which hides the recording and nothing else. You can delete an individual recording yourself at any time, locked or not, and deleting your account removes everything (see below).

The sweep is about media, so it does not touch your answer to the “what are you preparing for” question — that is a few words, not a file, and it is kept until you change it or delete your account. You can clear it back to unanswered in your settings at any time.

• Recordings live in private storage buckets. Every read is authorized per-user by database row-level security — there is no public URL for any recording.
• Playback uses short-lived signed links that expire after five minutes.
• Everything is encrypted in transit over HTTPS.

Whoever and wherever you are, you can:

• Change your retention window, or delete any individual recording, at any time.
• Delete your entire account, in the app or at stagemirror.ai/account/delete.
• Ask us what we hold about you, or ask for a copy, by emailing privacy@stagemirror.ai.

If you are in the UK or the EEA, the GDPR also gives you the following rights, and we will answer any of them within one month:

• Access — a copy of the personal data we hold about you.
• Rectification — have anything inaccurate corrected.
• Erasure — have it deleted, which the account deletion above does immediately and in full.
• Restriction — have us stop using it while a dispute about it is resolved.
• Portability — receive the data you gave us in a machine-readable form, or have us send it to someone else.
• Objection — object to anything we do on the basis of our legitimate interests, including the product analytics named above.
• Withdraw consent — for anything we do on the basis of consent, at any time.

Exercising any of these is free and we will not treat you differently for it. If you think we have got it wrong you can complain to your data protection authority — in the UK the Information Commissioner’s Office, and in the EEA the supervisory authority for the country you live in. We would rather you told us first, at privacy@stagemirror.ai, but you do not have to.

Nothing here is a statutory requirement — no law obliges you to give us any of this. Your email address and a recording are required to use StageMirror, in the sense that there is no product without them; every other field, including your display name and what you are preparing for, is optional and the app works without it.

Your account data — email, display name, settings, session records and scores — is kept for as long as the account exists, and goes when you delete it. Recordings go sooner, on the schedule above.

The scores and coaching are generated automatically, but they are feedback on a practice take: they do not decide anything about you, and nothing in StageMirror produces a legal or similarly significant effect within the meaning of Article 22.

You must be 13 or older to use StageMirror. We do not knowingly collect information from children under 13.

If we change how any of this works we will update this page and, for anything material, tell you in the app before it takes effect.