1. Who we are (data controller)

The data controller for personal data processed through the Service is:

• LIPITT SAS, a French simplified joint-stock company

• Registered office: 7 avenue de Laponie, 91940 Les Ulis, France

• SIREN/SIRET: 980 765 531 00010

• Privacy contact: privacy-parla@equos.ai

• Data protection point of contact: Antoine Ménager, CEO

For any question concerning your personal data or the exercise of your rights, please contact us at the email address above.

2. What data we collect

2.1 Account data (collected when you sign up)

You create an account using Sign in with Apple. From Apple, we receive:

• Your email address (which may be a private relay address provided by Apple if you choose to hide your real email)

• Your first name (as provided by you)

• A unique Apple user identifier specific to our app

2.2 Learning profile data (collected during onboarding)

To personalise your learning experience, we ask you to provide:

• Your current English level (e.g., beginner, intermediate, advanced)

• Your main objective for improving your English

• The area you most want to improve (e.g., conversation, business, travel)

2.3 Conversation data

• Voice (audio): streamed in real time during your conversations with the in-app avatar

• Video (camera): only if you choose to enable the camera; used to give the avatar contextual feedback on your facial expressions

• Conversation memory summaries: short text summaries of past conversations that we generate to give the avatar continuity (e.g., topics covered, vocabulary practiced). The full audio of your conversations is never stored by Lipitt

See Section 3 for important details on how voice and camera data are processed.

2.4 Subscription and payment data

Payments are processed exclusively through Apple's In-App Purchase system, managed by our subscription infrastructure provider RevenueCat, Inc. Lipitt never receives or stores your payment card details. We receive from Apple/RevenueCat only the subscription status (active, expired, in trial, cancelled) linked to a transaction identifier.

2.5 Technical and usage data

When you use the App, we automatically collect:

• Device information: device model, iOS version, language and region settings, time zone

• App-specific identifiers (Apple's IDFV — Identifier for Vendors)

• Usage analytics: screens viewed, features used, session duration, conversation initiation events

• Crash and performance logs (when the app crashes or encounters an error)

• IP address (used transiently for connection routing and abuse prevention)

We do not use Apple's IDFA, do not request App Tracking Transparency permission, and do not track you across other apps or websites.

2.6 Communications

If you contact us by email or interact with us, we keep a record of that correspondence.

3. Voice and camera data — special notice

Important. Audio and (optional) video data may qualify as "biometric data" under GDPR when processed in certain ways. We process your voice and camera input solely to enable the conversational learning experience, and never to uniquely identify you as an individual. We do not perform voice cloning, face recognition, or speaker identification.

3.1 How real-time conversations work

When you start a conversation:

1. Your microphone audio (and, only if you have enabled the camera, your video frames) are streamed in real time from your device to Google's Gemini Live API (operated by Google Ireland Limited, processing region: europe-west).

2. Gemini transcribes your speech, generates a response, and streams synthesised speech back to your device, all in real time.

3. The text response is also used by our on-device avatar engine (running locally on your iPhone) to animate the avatar's lip movements and facial expressions. No video of the avatar leaves your device.

3.2 Storage by Lipitt

Lipitt does not record, store or archive the raw audio or video of your conversations on its own servers. Only short text summaries (used as long-term "memory" for the avatar) are stored in our backend.

3.3 Storage by Google (Gemini Live API)

Google processes the streams as our sub-processor under its Cloud Data Processing Addendum. Based on Google's published policies:

• We have opted out of using your data for model training. Google does not use your prompts or responses to train its AI models.

• Google may temporarily log prompts and responses for a limited period for the sole purpose of detecting abuse and violations of its acceptable use policy.

• If session-resumption features are used, conversation state may be cached for up to 24 hours to allow reconnection; Lipitt does not enable session resumption beyond what is strictly necessary for service continuity.

For full details, see Google's Cloud Data Processing Addendum and the Gemini API data retention documentation.

3.4 Camera is optional and revocable

The camera permission is not required to use Parla. You can grant it, refuse it, or revoke it at any time in iOS Settings → Parla → Camera. The microphone is required for the core conversational feature; you can disable it in iOS Settings → Parla → Microphone, but this will prevent the conversational feature from working.

4. How we use your data and legal bases

Under the GDPR, we must rely on a "legal basis" to process your personal data. The table below sets out our processing purposes and the corresponding legal basis.