PRIVACY

Soulfi App - Microsoft Azure OpenAI (incl. the Whisper model)

Version date: 2026-06-04

Important note: This notice explains how the Soulfi App processes personal data for AI chat and optional voice input features. It is intended as a clear, user-friendly legal notice for international use.

1. Data controller

Soulfi GmbH

Email: privacy@soulfi.com

Phone: +49 40 22860138

Data Protection Officer: Maike Kruse

Email: privacy@soulfi.com

2. What this notice covers

This notice explains data processing for:

AI chat,

optional voice input (microphone),

storage and deletion of chat content.

Voice input is optional. You can always use chat by typing.

3. AI chat with Microsoft Azure OpenAI

3.1 What is this service?

Soulfi uses Microsoft Azure OpenAI Service to generate AI-based chat responses.

3.2 How an AI request works

You enter a message.

Example: "I feel anxious about tomorrow."

• Your message is transmitted in encrypted form (for example, TLS).
• The service processes your message and generates a response.
• The response is shown in the app.
• Example: "That is understandable. Here are three short breathing techniques that may help now."

3.3 Practical text-chat example

Typical flow:

You type: "I feel very nervous today."

• The message is securely transmitted.
• The service generates a suitable response.
• The response appears in your chat history.

3.4 AI model training

Under the contractual terms used by Soulfi with Microsoft Azure OpenAI, customer data are not used to train base models.

3.5 Processing location

AI processing (chat and speech-to-text) is performed on Microsoft Azure infrastructure in the EU region Sweden (Sweden Central), based on the configured setup.

Soulfi's app backend (for example account and chat data) is hosted on servers of Hetzner Online GmbH in Germany.

Processing in third countries outside the EU/EEA is not intended under the configured setup.

4. Voice input and speech-to-text (Whisper via Microsoft Azure OpenAI)

4.1 What happens during voice input?

You can optionally speak instead of typing. Your speech is converted into text and that text is processed in the chat.

4.2 Processing flow

Audio is recorded locally on your device.

Example: You say: "I am anxious about tomorrow's appointment."

• Transmission starts only after active user action (for example, tapping microphone/send).
• Audio is converted into text via OpenAI's Whisper model, operated through Microsoft Azure OpenAI.
• Example: Speech becomes text: "I am anxious about tomorrow's appointment."
• The transcribed text is processed in the chat context.

4.3 Practical voice-input example

You press the microphone button.

You speak your message.

• Audio is converted into text.
• The text is processed like a normal chat message.
• You receive the AI response in chat.

4.4 Text input vs. voice input

Text input (default):

You type a message.

• The text is processed.
• The response appears in chat.
• Voice input (optional):
• You speak a message.
• Audio is converted into text.
• The text is processed exactly like normal chat input.

Both paths lead to the same functional result. Voice input is an optional convenience feature.

4.5 Audio and text data

Audio data are processed for speech recognition.

The transcribed text is relevant for chat history.

Voice input can be disabled via operating system permissions.

5. Legal bases

Where the GDPR applies, processing is based in particular on:

• Art. 6(1)(b) GDPR (performance of contract) for core chat features,
• Art. 6(1)(a) GDPR (consent) for optional microphone use,
• Art. 9(2)(a) GDPR (explicit consent), where special category data are provided.

Outside the GDPR scope, equivalent legal bases under applicable local law may apply.

6. Retention and deletion

Chat content is stored according to product-side retention rules (currently up to 12 months), unless deleted earlier.

Users can delete chat content in the app.

Audio data from voice input are retained only as needed for technical processing and then deleted according to active system rules.

Practical examples:

Individual messages can be removed.

Full chat history can be deleted in app settings.

Microphone permissions can be withdrawn at any time in the operating system.

7. Recipients and processors

To provide the service, Soulfi uses technically required service providers, including Microsoft (Azure OpenAI, including the Whisper model for speech-to-text).

Relevant data-processing agreements are in place where legally required.

8. Security measures

Soulfi applies appropriate technical and organizational safeguards, including:

transport encryption,

access controls,

role-based permissions,

security and availability controls in cloud infrastructure.

9. Your rights

Subject to applicable law, you may have the right to:

• access,
• rectification,
• erasure,
• restriction of processing,
• data portability,
• objection,
• withdrawal of consent with future effect.

Please send requests to: Datenschutz@soulfi.de

10. Complaints

You may lodge a complaint with your local data protection authority, where applicable.

For users in the EU/EEA, this is typically your local supervisory authority.

11. Sources and further information

Microsoft Privacy: https://aka.ms/privacy

Microsoft Trust Center: https://azure.microsoft.com/support/trust-center/

Azure OpenAI: https://learn.microsoft.com/azure/ai-services/openai/

Azure OpenAI – Whisper (speech-to-text): https://learn.microsoft.com/azure/ai-services/openai/whisper-quickstart

12. Signing in to the app

Using the app requires a user account. You can sign in with your email address and a password. On Apple devices you can additionally use "Sign in with Apple"; in that case Apple shares your name and email address or, at your choice, an anonymized Apple relay email address. Sign-in via other third parties (such as Google) is not offered.

13. Location-based feature (weather)

The app can optionally use your approximate location ("While Using the App") to show your local weather and a matching daily greeting on the home screen. For this, your approximate coordinates are sent to the weather service Open-Meteo (open-meteo.com), which returns the current weather data. If location access is not granted, your approximate location may instead be derived from your IP address (service: ipwho.is). The legal basis is your consent (Art. 6(1)(a) GDPR). You can withdraw the location permission at any time in your operating system settings.

14. Changes to this notice

Soulfi may update this notice due to legal, technical, or product-related changes. The latest published version in the app or related information pages is authoritative.