Xentra

    AI Data Processing at Xentra

    Transparency notice on the use of artificial intelligence: which Xentra features involve AI, which provider receives the data, and what is redacted before it is sent.

    Last updated: 3 October 2026

    1. Overview

    Xentra uses two providers for AI-assisted features: Google LLC (Gemini API) and Anthropic PBC (Claude API), both based in the USA. This notice explains which feature transmits which data to which provider and what safeguards apply. Transfers take place on the basis of the European Commission's standard contractual clauses.

    Many AI features are optional and can be disabled in the settings. Translation of incoming messages cannot currently be switched off.

    2. AI Functions in Detail

    These features access an AI language model. The "Redacted" column indicates whether personal details are removed or replaced before transmission.

    FunctionProviderData sentRedacted
    Automatic guest replyGoogle GeminiMessage history, booking data (anonymized)✅
    AI assistantGoogle GeminiUser questions, booking context, preview of the latest guest message in waiting conversations (only when real guest names are enabled)⚙️ Configurable
    Onboarding assistantAnthropic ClaudeConversation, uploaded property documents (pseudonymised, credentials removed)✅
    Meeting summaryAnthropic ClaudeTranscript of recorded calls with us (only with consent)—
    FAQ generatorGoogle GeminiGuest messages (anonymized)✅
    Guest page generatorGoogle GeminiAccommodation info, past messages (anonymized)✅
    Property description and website copyGoogle GeminiProperty, amenity and profile details of the host—
    Invoice extractionGoogle GeminiInvoice PDF (host business document)⚠️ *
    Text improvementGoogle GeminiUser-entered text—
    Automatic translationGoogle GeminiText of the message being translated; e-mail addresses, phone numbers, links and known names (guest, host) as placeholders only✅
    Voice note transcriptionGoogle GeminiAudio content of the voice note—

    * Invoice PDFs are host business documents and are under the host's responsibility.

    3. Anonymization Measures

    For some guest-facing features, personal details are automatically redacted before transmission:

    • Last names are removed
    • Email addresses are replaced with [EMAIL]
    • Phone numbers are replaced with [PHONE]
    • UUIDs are removed

    The host-facing AI assistant uses the real guest names from your own bookings by default, so questions about guests can be answered directly. Google processes this data solely to answer the request (no training on API data). In Settings (Security → AI & privacy) you can switch to pseudonymization ("Guest 1", "Guest 2", …) instead.

    This redaction does not apply equally everywhere. When a message is translated, e-mail addresses, phone numbers, links and the names known to us (the booking's guest, the host) are replaced by placeholders and put back after translation; other details in the text, such as further names or addresses, are transmitted as written. Transcription requires the full audio, and invoice extraction requires the full document. The table above states for each feature whether redaction takes place.

    4. Audit Logging

    Each AI call is logged in an audit table. The following metadata is captured:

    • Name of the called function
    • Categories of data sent
    • Whether anonymization was applied
    • Timestamp and user ID

    No prompt contents or responses are stored in the audit log — only metadata.

    5. Consent Mechanism

    The automatic guest reply requires explicit consent to AI data processing in the bot settings.

    Without this consent, the bot does not transmit any data to Google Gemini. Other features, in particular the translation of incoming messages, do not depend on this consent.

    Disclosure to guests

    If a host answers guest messages fully automatically, the guest must be able to recognise that they are interacting with an AI system — as required by Article 50 of the AI Regulation (EU) 2024/1689. Xentra provides a disclosure notice for this, which the host activates in the settings. Responsibility for the disclosure lies with the host, because the message is sent in their name.

    6. No Training by Google

    Under the terms of use of both providers, data sent via the application programming interface is not used to train their models.

    7. Legal Basis

    Processing is based on consent (Art. 6(1)(a) GDPR) and legitimate interest (Art. 6(1)(f) GDPR).

    8. Contact

    For questions about AI data processing, please contact:

    Stellar Tourism Innovations GmbH
    E-Mail: datenschutz@myxentra.com

    We use cookies

    This website uses cookies and similar technologies for analytics and to improve your experience. Learn more in our Privacy Policy.