Usage data (e.g., visited websites, interest in content)
Meta-/communication data (e.g., device information, IP addresses)
Chat histories and interactions with our AI systems
Voice data (microphone audio while a session is running) and its transcripts when you use our voice features
Device and pairing data from our apps (device name, device token)
Credentials and content of services you connect as a connector (Section 17)
Which of these the Sales Companion iOS app and the Hy app process in detail is described in Section 15 and Section 16.
3. Purpose of Processing
We process your personal data for the following purposes:
Providing our AI chat services
Improving our services
Communicating with users
Fulfilling legal obligations
Ensuring IT security
4. Legal Bases
The processing of your data is based on the following legal grounds:
Consent (Art. 6(1)(a) GDPR)
Contract fulfillment (Art. 6(1)(b) GDPR)
Legal obligations (Art. 6(1)(c) GDPR)
Legitimate interests (Art. 6(1)(f) GDPR)
5. Recipients of Data
Your data may be shared with the following recipients:
Hosting providers (Hetzner Cloud, Hetzner Object Storage with Data-Processing Agreement)
Email provider (Strato with Data-Processing Agreement)
AI service providers (Anthropic, Google, OpenAI, Mistral, Perplexity, Melious AI)
IT service providers (WordPress, Brevo, ManyChat, N8N)
Payment service providers (Stripe, PayPal)
Clients/end-users of the chat service on their websites
Apple (speech recognition for the dictation feature of our iOS app, see Section 15; Siri shortcuts and notifications of the Hy app, see Section 16)
Service providers for error analysis and quality assurance (Sentry, Langfuse)
Providers of the services you connect as a connector, to the extent your assistant retrieves or creates something there on your behalf (Section 17)
Which additional recipients are involved when you use the Sales Companion iOS app or the Hy app is described in Section 15 and Section 16.
5a. Data Processing Agreements with AI Service Providers
To protect the data generated and used by our chat-bot customers, we have concluded Data Processing Agreements with our AI service providers. These agreements require the providers to:
Not use data for training their models (contractually excluded)
Delete data according to defined retention policies
Process data only on our instructions
Protect data with appropriate security measures
These agreements cover all content we transmit to the providers – expressly including real-time audio and its transcription through the voice features of our applications (OpenAI, model gpt-realtime).
For detailed information on data transfers to the United States and compensatory protective measures, see our Data Processing Agreement (DPA), available in your account under "Legal".
6. Retention Period
We store your personal data only for as long as necessary for the purposes mentioned or as required by legal retention periods:
Chat histories and logs: Maximum 90 days, then automatic deletion (Art. 5 GDPR – storage limitation principle)
File uploads: Maximum 90 days, then automatic removal from active systems
Backup archives: Longer retention for disaster recovery purposes (up to 12 months for database backups)
Voice data: Audio is not stored; it is relayed only for the duration of the session
Assistant memory: Content the assistant remembers permanently is retained until you have it deleted or until your account is deleted
Device pairing: Pairing codes expire after 10 minutes; device tokens are retained until revoked or until your account is deleted
The retention periods that apply to the individual features of the Sales Companion iOS app and the Hy app are described in Section 15 and Section 16.
7. Your Rights
You have the following rights:
Access to stored data
Correction of inaccurate data
Deletion of data
Restriction of data processing
Data portability
Objection to processing
Withdrawal of given consents
Lodging a complaint with a supervisory authority
8. Cookies and Tracking
We use cookies and similar technologies (Google Analytics and PostHog, each with opt-in) for the following purposes:
Website functionality
User behavior analysis
Improving our services
After your consent, PostHog stores both a cookie and an entry in your browser's localStorage (each prefixed with ph_) in order to remember your consent and identification state across page loads. These are removed when you withdraw your consent.
9. Data Security
We implement technical and organizational security measures to protect your personal data against accidental or intentional manipulation, loss, destruction, or unauthorized access.
10. AI-Specific Processing and Data Protection
Our AI chatbots process your inputs to generate personalized responses. This processing is:
Automated by AI systems from Anthropic, Google, OpenAI and other providers
Based on chat history (retained for a maximum of 90 days)
Aimed at providing the best possible answers to your inquiries
Optionally with PII masking: Upon request, we mask sensitive data (email addresses, phone numbers, names) before transmission to AI providers. You see the real data, the provider sees only placeholders
11. Embedding on Third-Party Websites
Our chat widget can be embedded by third parties (our clients) on their websites. In such cases:
Data processing on our clients' websites is their responsibility and subject to their privacy policies.
We have no direct influence over the type and scope of data processing on our clients' websites.
Our clients are obligated by our terms and conditions to:
Comply with applicable data protection regulations
Inform their users about data processing through our widget
Obtain necessary consents, if applicable
For questions regarding data processing on our clients' websites, please contact the respective website operator directly.
12. Use of Google OAuth SSO
We offer the option to log in using your Google account via the OAuth Single Sign-On (SSO) service. This service is provided by Google Ireland Limited, Gordon House, Barrow Street, Dublin 4, Ireland.
As part of the authentication process, the following data is retrieved from your Google account:
Your email address
The processing of this data is carried out to simplify the login process and provide our services. The legal basis for this processing is Article 6(1)(b) GDPR (performance of a contract).
Google acts as an independent data controller for the processing of your data within the OAuth process. For more information on how Google processes data, please refer to their Privacy Policy.
If you wish to disconnect your Google account from our service, you can do so at any time in your Google account settings.
13. Use of Google Analytics
Our website uses Google Analytics, a web analytics service provided by Google Ireland Limited, Gordon House, Barrow Street, Dublin 4, Ireland. Google Analytics uses cookies to analyze how you use the website.
The following data is collected:
IP address (anonymized if IP anonymization is enabled)
Visited pages and time spent
Location data (based on IP address)
Device type and browser
The processing of this data is based on your consent (Article 6(1)(a) GDPR). The collected data is used to evaluate the use of our website and improve our services.
You can prevent the collection of your data by Google Analytics by installing the opt-out browser add-on provided by Google. For more information, please refer to Google's Privacy Policy.
14. Use of PostHog (Product Analytics and Session Replay)
Our application uses PostHog, a web and product analytics service provided by PostHog, Inc., 2261 Market St., #4008, San Francisco, CA 94114, USA. We operate the EU instance of PostHog (eu.i.posthog.com); your usage data is processed and stored on servers located in Frankfurt am Main, Germany.
Transfer to a third country: although the data is stored within the EU, PostHog, Inc. is a US-based company. We have concluded the EU Standard Contractual Clauses (Article 46(2)(c) GDPR) and a Data Processing Agreement (Article 28 GDPR) with PostHog to ensure an adequate level of data protection within the meaning of the GDPR.
The following data is collected:
IP address: on the EU instance of PostHog, IP capture is disabled by default for all new projects. Where it is enabled in the project settings on an exceptional basis, it is used only for geolocation and abuse prevention.
Pages visited (including the full URL and any query parameters), clicks, and time spent
Device type, operating system, and browser
An anonymous device identifier stored in a cookie and in your browser's localStorage
After you sign in: your user ID, so that usage events can be associated with your account
After consent and only in screen recordings (Session Replay): layout interactions such as clicks, mouse movements, scrolling, and navigation. All input fields and all rendered text are masked before recording — passwords, chat content, email addresses, and other personal text content are never captured. The visited page URLs (including query parameters) are recorded. Console logs and network metadata are explicitly not captured. The chat widget embedded on customer websites is entirely excluded from Session Replay.
Processing is based on your consent (Article 6(1)(a) GDPR) and is only activated after you accept the cookie banner. Before consent, no cookies or localStorage entries are set for PostHog and no data is transmitted. If you decline consent or withdraw it, capture stops immediately and stored identifiers are cleared locally.
You can withdraw your consent at any time via the cookie settings. For signed-in users, deleting your account additionally triggers deletion of the associated PostHog person record including all linked events (Article 17 GDPR).
Note for non-signed-in visitors: prior to sign-in, your PostHog person identifier is generated locally in your browser and is not known to us in plaintext. To satisfy an access or erasure request from a non-signed-in visitor we need the PostHog person identifier stored in your browser (readable via posthog.get_distinct_id() in the browser console) or another identifying detail.
The preceding sections apply to our "Sales Companion" iOS app accordingly; the following describes what is specific to the app. The controller is HybridAI GmbH as named in Section 1.
15.1 Voice Features (Microphone Access)
The app uses the microphone for live questions to the assistant, for roleplay training, and for listening in during coaching (see Section 15.2). Access only takes place if you grant it in iOS and actively start a voice feature.
Transmission path: While a session is running, the audio is streamed in real time to our servers in Germany and relayed from there to OpenAI (model gpt-realtime, processed in the United States), see Sections 5 and 5a.
No audio storage: We relay the audio stream only. No recordings or audio files are created on our servers. All that is retained are the session's usage figures (model, duration, tokens) for billing purposes.
Transcripts: Speech recognition converts what is spoken into text.
For live questions to the assistant, the text is treated like a chat message and becomes part of your chat history – retained for a maximum of 90 days (Section 6).
For roleplay and listening in, the transcript is not stored – neither in your chat history nor anywhere else in our systems. It is nonetheless processed: on your device during the session, on our servers, by the language model, and once more for the post-call review (Section 15.2). Once each of those steps is finished, it remains only on your device.
15.2 Listening In During Coaching and the Data of the People You Speak With
When listening in, the assistant accompanies a real conversation and picks up, through your device's microphone, the voices and statements of the people you are speaking with – usually your customers. A single microphone channel is transmitted, without speaker attribution.
Your obligation as a user: You may only use the listening-in feature if you have informed everyone taking part in the conversation beforehand and obtained their consent. Listening in or recording covertly is unlawful and may be a criminal offence (Section 201 of the German Criminal Code). If any participant objects, you must not use the feature.
Legal basis for third-party data: You, or the company you work for, decide whether to use the feature; to that extent you or your company are the controller for the processing of the data of the people you speak with. We process that data on your instructions as a processor under a Data Processing Agreement (Art. 28 GDPR). The legal basis towards the people you speak with is their consent (Art. 6(1)(a) GDPR), which you obtain.
Storage: The transcript of the conversation stays on your device while the session is running. For the post-call review, the app transmits the transcript once, together with your briefing and the prompts shown to you during the call, to our server and to the language model used. The review is generated and returned to your device; neither the transcript nor the review is stored by us. The finished review is then held only on your device.
15.3 Text and Context Data
Chat messages you send to the assistant.
Briefings you enter before a call or roleplay (e.g., customer name, context, objective) – these may contain personal data of third parties; Section 15.2 applies accordingly.
Roleplay transcripts you submit for review.
Context sent automatically: So the assistant can refer to where you currently stand, the app automatically sends your list of open tasks along as context whenever you make a request.
This content is stored as part of your chat history and deleted after 90 days at the latest (Section 6). Where the assistant remembers details permanently at your request (memory feature), those details are retained beyond the 90 days until you have them deleted or your account is deleted.
15.4 Account, Device Pairing and Camera
Account: We process your email address and your user ID in order to associate the app with your account. The legal basis is Art. 6(1)(b) GDPR (contract fulfillment).
Camera: The app uses the camera solely to read the QR code during device pairing. The code is evaluated on the device; no images or video are transmitted to us or stored.
Pairing code: The QR code shown in the web interface is valid for 10 minutes and can be redeemed once. We store it only as a hash value; spent codes are removed automatically.
Device token and device name: Once pairing succeeds, the device receives its own access token, which we store only as a hash value. We additionally store the device name you chose (e.g., "Anna's iPhone", which may identify a person), the time of pairing and the time of last use. You can revoke each paired device individually in the web interface under "Connected devices"; the token stops working immediately.
15.5 Dictation (Apple)
When you dictate tasks, the app uses the operating system's speech recognition. Depending on your device and settings, Apple processes the recording on the device or on servers operated by Apple Inc. or Apple Distribution International Ltd. Apple acts as an independent controller in this respect; for details see Apple's Privacy Policy. If you would rather avoid this, please enter tasks using the keyboard.
15.6 Competitor Monitoring ("Watch")
This feature displays notes your agent previously stored in its cloud memory. These are predominantly company data; the notes may, however, also contain information about individuals (such as contact persons or publicly acting representatives of the companies being monitored). The content is retained until you delete it from the agent's memory.
15.7 Data Stored Only on the Device
The following data stays on your device and is not transmitted to us:
Your tasks, saved answers and the activity data behind the progress rings – each kept separately per account.
Credentials and the device token, which are held in the iOS keychain.
Notifications are generated locally on the device only; the app does not send push notifications through any external service.
This data is deleted when you remove the app from your device.
15.8 No Tracking in the App
The app contains no analytics or tracking tools: no PostHog, no Google Analytics, no advertising or device identifiers used for advertising purposes, no product analytics and no crash reporting. It bundles no third-party libraries at all. This is distinct from the records required for billing and operations: for every AI request we retain the model, the time and the consumption (Section 15.1); those records describe usage, not your behaviour in the app. Sections 13 and 14 concern our website and the browser-based workspace only and do not apply to the app.
15.9 Recipients and Transfers to Third Countries
OpenAI (United States) for real-time speech and its transcription – safeguarded by a Data Processing Agreement and Standard Contractual Clauses, see Section 5a.
The model provider of the bot in use for text requests and reviews, see Sections 5 and 5a.
Apple for the speech recognition behind the dictation feature, see Section 15.5.
15.10 Legal Bases, Consent and Withdrawal
AI processing (voice features, chat, reviews): your consent under Art. 6(1)(a) GDPR, which the app requests before first use. You can withdraw your consent at any time with effect for the future – in the app under Settings → Privacy → "AI processing". After withdrawal, the app's AI features are no longer available; the lawfulness of processing carried out up to that point is unaffected.
Account and device pairing: Art. 6(1)(b) GDPR (contract fulfillment).
15.11 Deleting and Disconnecting
Disconnecting a device: In the app via "Disconnect device" or "Sign out"; in addition, you can revoke the device in the web interface under "Connected devices".
Local data: The data stored only on the device (Section 15.7) is deleted when you remove the app from your device.
Server-side data: The retention periods in Section 6 and your rights under Section 7 apply to chat histories and all other data.
16. Hy App (iOS and Android)
The preceding sections apply to our "Hy" app for iOS and Android accordingly; the following describes what is specific to the app. The controller is HybridAI GmbH as named in Section 1.
16.1 Voice Calls, Siri and Shortcuts
The iOS app uses the microphone for calls with Hy after you grant microphone access and start or accept a call. Calls from Hy require separate activation in Settings. An outgoing call normally ends when you leave the app; calls through CarPlay and accepted incoming calls can continue in the background or while the phone is locked. You can end the call using the call controls. The Android app has no voice features.
Transmission path: While a call is running, the audio is streamed in real time to our servers in Germany and relayed from there to OpenAI (model gpt-realtime, processed in the United States), see Sections 5 and 5a.
No audio storage on our servers: We relay the audio stream without creating recordings or audio files on our servers. This does not mean that no call data is retained.
Transcripts and usage: The text of both sides becomes part of your chat with Hy and follows the retention rules in Section 16.2. We also retain the call's usage figures (model, duration, tokens) for billing and service operation.
Optional Siri and Shortcuts: “Ask Hy” receives your question as text from Siri or Shortcuts and sends it through Hy's normal chat processing. The question and response are stored with the conversation (Section 16.2). Hy returns the answer text to Siri to read aloud, or to Shortcuts as the action's result. The answer can include personal information from your conversation and connected sources, including Health and Contacts. “Talk to Hy” opens Hy and starts the call flow described above; you can also assign this shortcut to the Action Button. Both shortcuts require device authentication. Hy's sign-in and AI-consent requirements still apply, and calls require microphone permission.
Apple's Siri processing: Apple processes Siri interactions independently under its Siri privacy notice. Processing depends on your device and settings; Apple may retain transcripts and use them to improve Siri. Our training restrictions in Section 16.5 do not govern Apple's independent processing. You can use Hy directly without Siri.
16.2 Text and Context Data
Chat messages you send to Hy, and your todos and goals.
Photos, camera pictures and files you attach. Before sending, the app reduces photos to at most 2048 pixels and removes their metadata, including location data.
Context sent automatically: So that Hy can refer to your day, the app sends data from the sources you connect in the app:
Calendar: on iOS, appointments from the past 30 days to 365 days ahead; on Android, the next 7 days. This includes titles, start/end times, locations, attendee names and, on iOS, calendar names and sources.
Reminders (iOS only): open reminders that are overdue or due within the next 7 days (title and due time).
Health (Apple Health on iOS, Health Connect on Android): daily totals for the past 7 days (steps, exercise minutes, active energy, resting heart rate, sleep duration) and your workouts. The app reads this data on your device and transmits only these summaries, no individual measurements.
Contacts (iOS only): names, nicknames, companies, job titles, email addresses, phone numbers, birthdays and related names from the address book you authorize. You can limit the selection in iOS. The app only reads contacts.
Hy keeps this content, its memory and your files in your Hy workspace on our servers in Germany. Unlike the chat histories in Section 6, they are not deleted automatically after 90 days; they remain until you have them deleted or your account is deleted. Backups are subject to Section 6. If another person gives you access to their agents, for example as a member of their team, what you write to those agents is stored in that person's Hy workspace, not in yours. It belongs to their workspace and is deleted with it (when they have it deleted or their account is deleted), not when your account is deleted.
Video analysis (iOS only): when you attach a video, the phone measures body movement and sends the measurements, analysis notes and selected still images to Hy. The video, sound and detailed pose track remain on the phone. These materials are used to answer your request and are stored with the conversation and files described above.
16.3 Account, Device Pairing and Camera
Account: You sign in with your HybridAI account in your device's browser; the app never sees your password. We process your email address and your user ID in order to associate the app with your account. The app receives its own access tokens, which we store only as hash values. The legal basis is Art. 6(1)(b) GDPR (contract fulfillment).
Camera: The app uses the camera for pictures you attach (Section 16.2) and to read the QR code during device pairing. On iOS the code is evaluated on the device; on Android it is read by the code scanner of Google Play services, and the app itself has no camera access. No images of the code are transmitted to us or stored.
Pairing code, device token and device name: Section 15.4 applies.
16.4 Data Stored on the Device
The following data is kept on your device:
Copies of your conversations and of the files you sent, in protected app storage that is excluded from backups.
Credentials and access tokens, held in the iOS keychain or encrypted with a key from the Android Keystore.
Signing out removes this data from the device. Uninstalling removes app storage; iOS Keychain entries may remain, so sign out before uninstalling to remove saved credentials.
16.5 Personalization, Reports and No App Analytics
Hy does not integrate advertising, product-analytics or crash-reporting SDKs into its native app. We do not use its data for advertising tracking. The native iOS app bundles no third-party libraries; Android uses Google's Android libraries (including Health Connect and the Google Play services code scanner) and OkHttp. This does not describe external web content opened or loaded by the app; see Section 16.10. We retain the model, time and consumption of AI requests for billing and service operation (Section 16.1). The website analytics in Sections 13 and 14 concern our website and browser workspace, not native Hy screens. The following personal preferences serve your own recommendations, not aggregate product analytics.
Personal suggestions: Hy uses relevant conversations, memories, instructions and sources you have connected to tailor replies, ideas and recommendations to you. On iOS, likes and hidden items in “For you”, dismissed ideas and your feed instructions are synchronized to your Hy workspace with an item identifier, title or instruction, action and timestamp, linked to your account. They can be included in requests to the selected AI provider. Merely opening, reading or bookmarking an item is not recorded as a preference. These records are separate from chat emoji reactions and are not used for model training or general product improvement. They follow the workspace retention rules in Section 16.2; changing a preference replaces its current value, rather than erasing content already included in conversations. The AI consent in Section 16.7 also covers this personalization.
Emoji reactions: All emoji reactions, including thumbs up and thumbs down, are saved with the message in your Hy workspace and can be included in the context of your next conversation with Hy. They do not create answer ratings or submit feedback for product improvement.
Reports: When you choose Report, the selected answer, preceding message, reason and optional note are sent to HybridAI to investigate the reported problem. On iOS, the report also includes the Hy, conversation and message identifiers and app version/build, and is associated with your account. Reports are separate from reactions and are not used for model training or general product improvement. Retention follows Section 6.
Health and Contacts: For Hy, HybridAI excludes Health and Contacts data, including summaries and answers derived from it, from its training, evaluation and product-improvement pipelines. This restriction also applies after you disconnect the source. The data may still be processed to fulfil your requests; content you explicitly submit in a report is used only to investigate that report. Our contracted AI providers may not use this data to train their models. If you choose to use Siri, Apple's independent processing described in Section 16.1 also applies.
16.6 Recipients and Transfers to Third Countries
Melious AI GmbH (Germany) for EU model inference when this option is available and selected. Requests, relevant conversation context and responses pass through its inference gateway to European inference providers. Its subprocessor notice identifies those providers and their processing roles; its DPA describes the contractual terms. Changing the chat model does not change the separate voice or image provider.
OpenAI (United States) for real-time speech in calls and its transcription – safeguarded by a Data Processing Agreement and Standard Contractual Clauses, see Section 5a.
The model provider of the assistant in use for text requests, see Sections 5 and 5a.
Apple when you choose Siri: the question and returned answer are handled as described in Section 16.1. Apple processes Siri interactions under its own privacy terms, separately from our contracted AI providers.
Apple for notifications on iOS: If you allow them, the app registers with the Apple Push Notification service. Your Hy workspace stores the APNs device token, and a PushKit token for incoming calls when enabled, linked to your account. The platform relay stores only a hash of the token; the runtime needs the token itself to address your device. These identifiers serve notifications and calls, not advertising. A notification contains Hy's name and, depending on its kind, the text of a reminder, the title of a suggestion or a fixed notice that Hy has finished or needs you. The Android app has no notifications.
16.7 Legal Bases, Consent and Withdrawal
AI processing (calls, chat, ideas, suggestions): your consent under Art. 6(1)(a) GDPR, which the app requests before first use. You can withdraw your consent at any time with effect for the future – in the app under Settings → Privacy & data → "AI processing". After withdrawal, the app's AI features are no longer available; the lawfulness of processing carried out up to that point is unaffected.
Health data: your explicit consent under Art. 9(2)(a) GDPR, which you give by connecting Health in the app. You withdraw it by disconnecting Health in the app or by removing the app's access in Apple Health or Health Connect.
Account and device pairing: Art. 6(1)(b) GDPR (contract fulfillment).
iOS withdrawal: new requests, queued messages and uploads from the phone stop when you turn off AI processing. The app also requests that Hy’s cloud schedules pause and confirms the result; if it cannot reach the server, it shows a notice and retries on return. Already shared content remains under the retention rules above. Consent is tied to the signed-in account and processing destination; signing out clears the local agreement and device-source switches.
Withdrawing AI consent in Hy also blocks new Hy requests through Siri or Shortcuts. It does not change your Apple Siri settings or delete data Apple has already processed; manage those separately in the device's Siri settings.
16.8 Deleting and Disconnecting
Signing out: In the app via "Sign out". This revokes the app's access and ends notifications; on iOS it also removes the calendar, reminder and Health summaries the app stored in your Hy workspace. You can also revoke paired devices in the web interface under "Connected devices".
Local data: Signing out removes the local data described in Section 16.4. Uninstalling removes app storage but may leave iOS Keychain credentials.
Server-side data: Section 16.2 applies to Hy's chats, memory and files, and the retention periods in Section 6 apply to all other data. Your rights under Section 7 apply to all data. You can delete your account in the web interface under "Account" or in the iOS app under Settings → Delete account.
16.9 Saved Website Logins and Browser Takeover (iOS)
If you save a website login, the app transmits the site's hostname, username and password to your Hy runtime. Credentials are stored in its encrypted credential store in your workspace in Germany and injected into the login form only for the matching site. The password is not returned to the language model through the saved-login tool. This is separate from signing in to your HybridAI account. You can remove a login under Saved logins; signing out of the mobile app does not delete these server-side credentials or necessarily end the website's existing browser session. Account deletion removes the workspace; backup retention follows Section 6.
During browser takeover, steps are temporarily recorded in the cloud browser. Website images, the page address and title are sent from Hy's cloud browser to your phone; your taps, scrolling and keyboard input are sent back to operate it. The website receives the actions and entries you make and processes them under its own privacy terms. If you choose to remember the procedure, the recorded steps are saved as a routine in your workspace for later use by Hy, including relevant page addresses and non-secret input. Fields marked as passwords or verification codes are excluded from the saved routine. Routines and any content retained in chat or memory follow Section 16.2. These features provide the actions you request under Art. 6(1)(b) GDPR; AI processing remains subject to Section 16.7.
16.10 External Previews, Media and Generated Pages (iOS)
When Hy displays a link or a feed item, the phone may contact the linked website, a preview service or an image host to obtain a title, thumbnail or image before you tap the link. Video previews and players can use YouTube (Google), Vimeo, Instagram (Meta) or TikTok, depending on the link; playback starts when you choose it. The private Apps gallery displays HTML pages generated for you. These pages may load HTTPS images and scripts, styles or fonts from cdnjs.cloudflare.com, cdn.jsdelivr.net and unpkg.com.
Those requests disclose your IP address, the requested resource address and technical request information to the respective hosts. Embedded players and websites may process further data under their own privacy and cookie terms; a temporary web view does not prevent transmission. This is distinct from our AI providers and does not imply that every host stores data in Germany or follows our no-training commitments. Requests serve the previews and content you use under Art. 6(1)(b) GDPR; any processing requiring consent needs separate consent. Opening an external website is also subject to that site's privacy terms.
17. Connectors (Connected Services)
You can connect your assistants in HybridAI and in the Hy app to your own accounts with other services, such as Google Workspace, Microsoft 365, GitHub, Jira, X or an email mailbox (IMAP/CalDAV). This only happens when you connect a connector yourself. The controller is HybridAI GmbH as named in Section 1; the respective providers remain responsible for the processing within their own services.
17.1 What Data We Process
Credentials: When you connect, the provider gives us access tokens; for an email mailbox, we receive your address and your password or app password. We store them encrypted on our servers in Germany, together with the permissions you granted.
Content from the connected service: Your assistant retrieves data when a connector tool is needed to fulfil your request, for example to read emails, appointments or files. In Hy this also includes proactive checks and scheduled tasks you enable: they can run on the server while the app is closed, without a new chat message. You can turn off the task or proactive feature, or disconnect the source. Retrieved content is sent to the selected AI provider for processing (Sections 5 and 5a), subject to Hy's AI consent in Section 16.7.
Write actions: With Google Workspace and Microsoft 365, your assistant can only draft or send emails, create appointments or create files if you have additionally granted that permission. For every tool of a connector, you can also decide that your assistant must ask for your confirmation first, or must not use it at all.
17.2 Storage
We do not keep content from connected services in a separate store. It becomes part of the conversation with your assistant and is stored with it; the periods in Section 6, or for Hy in Section 16.2, apply. If your assistant saves content to its memory, the period for the assistant's memory applies (Section 6). Credentials remain stored until you disconnect the connector or delete your account (Section 17.5).
17.3 No Access by Our Staff, No Training
No access: Our staff do not read content from connected services. Requests in which connectors are available to your assistant are recorded in our tools for error analysis, quality assurance and usage analysis without any content, i.e. only model, time, usage and, where applicable, the type of error. Exceptions apply only if you have given your explicit consent (for example in a support case), if it is necessary to prevent abuse or for security reasons, or if we are legally obliged to do so.
No training: We do not use content from connected services to develop, improve or train AI or machine learning models. Ratings (thumbs up/down) of answers that may contain such content are not stored as training data. Our AI providers are not permitted to use the content to train their models either (Section 5a).
No transfer for other purposes: We do not sell this data and do not use it for advertising.
17.4 Google Workspace
With the Google Workspace connector, depending on the permissions you grant at Google, your assistant can:
Gmail: search and read emails; with additional permission, create drafts and send emails on your behalf.
Google Calendar: read your calendars and events; with additional permission, create events and invite attendees.
Google Drive: search and read files; with additional permission, create files. Write access is limited to files the connector created itself.
HybridAI's use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements.
Legal basis: Art. 6(1)(b) GDPR (performance of a contract), because you set up and use the feature yourself. Where data of third parties is processed, such as the senders of your emails or the attendees of your events, Art. 6(1)(f) GDPR; our legitimate interest is being able to answer your requests.
Disconnecting: You can disconnect any connector at any time on the "Connectors" page on the web or in the Hy app. We then delete the stored credentials and revoke access at the provider, where the provider supports this (Google does).
Deleting your account: When you delete your account, we delete all of your connector connections.
18. Changes
We reserve the right to adapt this privacy policy to comply with current legal requirements.
19. Contact
If you have questions about the processing of your personal data, you can contact us at any time:
