App and website

Privacy Policy

Last updated: 2 November 2026

1. General

This Privacy Policy informs you about the processing of your personal data in connection with:

The sahmo service provides AI-supported transcription and structured analysis of audio recordings. You can record your own thoughts or conversations, or upload existing audio recordings and voice messages. The analyses serve informational purposes and self-reflection. The app is initially not available as a browser application.

The first analysis is possible without registration. Personal data is processed in that case too.

For persons in the United States, the information in section 15 and our Consumer Health Data Privacy Policy additionally apply.

sahmo may only be used by persons aged 18 and over. If we learn that a minor is using sahmo, we block access and review the deletion of data processed without a lawful basis. Section 6.5 applies to information about other people in recordings.

Recordings, transcripts and analyses may contain personal data about you and other people, including special categories of personal data depending on the content. We explain the details in sections 3 and 6.3.

2. Controller

The controller for the processing of your personal data within the meaning of the General Data Protection Regulation (GDPR) is:

ailloy labs GmbH
Feuerbachstraße 3
65195 Wiesbaden
Germany

You can reach us with data protection questions at info@ailloy.io.

3. Categories of personal data

We process personal data that is necessary for providing and using the sahmo service, that you provide, or that arises through your use. This also applies to use without a registered account. Mandatory information is information without which the respective function cannot be provided, or can only be provided to a limited extent.

3.1 Account and contract data

When you register a user account, we process your email address and a display name that you choose freely. Your full legal name is not required. You can sign in directly with sahmo or through Google or Apple. In doing so, we process the account and authentication data required for the respective sign-in method.

Even when you use sahmo without registering, a pseudonymous user ID is used, and your recording and analysis are assigned to it. If you subsequently create a user account, this ID is linked to your account, and the associated content and analyses are assigned to that account. This data is therefore not anonymous.

To check eligibility, we process your confirmation that you are of age and, where the respective store provides it, its age category or age signal. We use this information to enforce the minimum age and to meet applicable legal requirements. The app evaluates the store’s signal only on your device; we do not store it.

3.2 Content and communication data

This includes in particular:

The content may contain data about you, other speakers, senders of voice messages and people who are merely mentioned. Depending on what is said, this may include special categories of personal data, for example information about health, sex life, sexual orientation, religious beliefs or political opinions. We explain the legal bases that apply to this in section 6.3.

3.3 Derived analysis data

We create structured analyses from the recordings and transcripts. Depending on the format (dialogue or monologue), the following information is processed in particular:

The analyses may contain original words, verbatim questions and summaries. Descriptions of people or bodies mentioned may also make them identifiable. The derived information may itself contain personal data and special categories of personal data. Section 6.3 applies accordingly. Fixed definition texts for the categories are provided independently of the respective recording.

For speaker separation, AssemblyAI temporarily creates mathematical representations of the voices, known as speaker embeddings. They are used to attribute contributions within the same recording to different speakers. According to AssemblyAI, they are not retained beyond the delivery of the transcript. We do not use them to establish a person’s identity or to recognise anyone across different recordings. The speaker labels remain part of the respective transcript. Details on retention are set out in section 8.1.

We do not derive feelings, intentions or personality traits from tone of voice, volume, speaking speed or other voice characteristics. Nor does sahmo derive emotions, intentions, motives or personality traits from the content that are not explicitly named. Statements about your own experience and about other people are captured and classified based on what is explicitly said.

Conversation quality is not rated through scores or comparisons with other recordings. Whether a statement is true is not checked. Recommendations or instructions for action are not created. These limits apply both to background processing and to the results displayed.

3.4 Analyses across several recordings

With a user account, you see your recordings in your history and can group completed recordings into collections that you name yourself. From two recordings in a collection onwards, we create an additional analysis from them. For this we process the name you gave the collection, the language, the transcripts and the known metrics of the recordings it contains, in chronological order.

Every observation refers to the recording and the passage in the transcript it is based on. Speaker labels apply only within their recording; a person is not derived from them across several recordings. Collections are not available without a user account.

3.5 Technical usage and log data

The following data arises in particular during use:

When it starts, the app checks via Expo (EAS Update) whether an updated version of its interface is available and downloads it if so. In doing so, your IP address, platform, app and runtime version and a technical installation identifier are transmitted to Expo, but no recording content or account data.

Audio content, transcripts and content-related analyses are not stored in technical usage, error or processing logs. The same applies to content-related inputs and outputs of the AI systems as well as passwords and access tokens. Where such information may be contained in technical error messages, it is removed before logging.

After a recording ends, the app calculates a few rounded loudness values for it on your device and stores them with the recording’s identifier in the usage events. They serve technical diagnostics, for example to find out why no speech was recognised in a recording. They contain no audio and no course over time and allow no conclusions about what was said. We do not derive emotions, intentions or personality traits from them.

The separate processing of content data for quality assurance based on consent is described in section 5.

3.6 Notifications

If you agree to be notified when processing is complete, we process a push token of your device and the language set in the app for the respective recording. The notification contains only the processing status and the identifier of the recording, no audio content, transcripts or analyses. Delivery takes place through Expo and Apple or Google and is not technically guaranteed. You can switch notifications off at any time in your device settings.

3.7 Subscription and billing data

You buy the paid Premium subscription through the Apple App Store or Google Play. We do not process payment data such as card or bank details; it stays with the respective store.

We process the state of your access as confirmed by the store, in particular the product purchased, the payment period, the store, the purchase environment, the start of the current term and the expiry date, as well as the number of analyses you have used in the current billing month. We need this data to unlock your access and to calculate the allowance. Section 14.3 describes how these details are managed through RevenueCat.

4. Use and assignment of content

You can use sahmo for monologues and dialogues. Content is assigned to the user ID or the user account through which the recording was created or uploaded.

4.1 Visibility of content

The fact that a person speaks in a recording or is mentioned in it does not give them access to the recording, the transcript or the analysis. There is no link between user accounts and no shared visibility of content arising from one. There is no function for sharing analyses.

We use appropriate technical and organisational measures, in particular role-based access restrictions and technical verification mechanisms, to protect the confidentiality of the content. Authorised access for providing the service and, subject to separate consent, for content quality assurance follows the purposes and access restrictions described in this Privacy Policy.

As with all complex IT systems, technical malfunctions or unintended processing cannot be ruled out entirely. In such a case, we take appropriate measures to investigate and remedy the issue and inform data subjects and supervisory authorities where legally required.

4.2 Assignment on registration

If you create a user account after using sahmo without registering, the content and analyses previously assigned to your pseudonymous user ID are linked to your account. This does not create any connection to the account of another speaker.

5. Purposes of processing

We process personal data for the following purposes:

We explain the legal bases in section 6 and in the respective sections about the services used.

6.1 Performance of a contract (Art. 6 (1) (b) GDPR)

Where this is necessary to provide the functions you request and to carry out your usage relationship, we process your personal data on the basis of Art. 6 (1) (b) GDPR. This concerns in particular:

For special categories of personal data, the requirements in section 6.3 additionally apply. The usage contract with you does not automatically establish a legal basis for processing the personal data of other people.

On the basis of consent, we process in particular audio content, transcripts and content-related analyses for the separate quality assurance and product improvement described in section 5, the data for notifications under section 3.6, and the data for marketing attribution with Adjust under section 14.2.

You can withdraw consent at any time with effect for the future. The lawfulness of processing carried out up to the withdrawal remains unaffected. Consent to content quality assurance is voluntary and separate from accepting the Terms of Use.

6.3 Special categories of personal data (Art. 9 (2) (a) GDPR)

Recordings, transcripts and analyses may contain special categories of personal data, for example information about health, sex life, sexual orientation, religious beliefs or political opinions. Processing them requires, in addition to a legal basis under Art. 6 GDPR, an exception under Art. 9 (2) GDPR.

For processing such information about yourself, we obtain your explicit consent under Art. 9 (2) (a) GDPR in a separate confirmation step when you first open the app, that is, before your first recording and first upload. It covers the transcription described, including speaker separation, the analysis and storage. Consent validly given continues to apply unchanged after registration; we do not ask for it again when you register.

You can withdraw it at any time with effect for the future at info@ailloy.io. The lawfulness of processing up to that point remains unaffected. Without the required consent, such content cannot be processed further.

Section 6.5 applies to data about other people. Using sensitive content additionally for content quality assurance requires explicit consent covering that purpose.

6.4 Legitimate interest (Art. 6 (1) (f) GDPR)

On the basis of our legitimate interests, we process in particular technical usage and log data to ensure the stability, security and correct functioning of our systems. Where permitted and not subject to consent, we also process pseudonymised technical usage data to develop and improve the service.

Our legitimate interests lie in providing the sahmo service securely, reliably and economically, and in developing it further. Processing on this basis takes place only where your interests or fundamental rights and freedoms do not override them. You can object on grounds relating to your particular situation (see section 10).

Content quality assurance using audio content, transcripts and analyses takes place exclusively on the basis of the separate consent under section 6.2 and, where applicable, 6.3. Genuinely anonymised data without any personal reference is not subject to the GDPR; merely replacing names with user IDs does not constitute anonymisation.

6.5 Other people in recordings

Recordings may contain personal data of other speakers, of senders of voice messages and of people who are merely mentioned. We are the controller for our processing.

We process ordinary personal data on the basis of Art. 6 (1) (f) GDPR, where this is necessary for the requested transcription and analysis and the interests or fundamental rights of the data subjects do not override it. Our interest and the user’s interest lie in preparing lawfully provided conversation content for personal reflection. Confidentiality, the expectations of the data subjects and the scope of the analysis must be taken into account in particular.

Sensitive information about other identifiable people, for example about their health or sex life, may only be processed with their own explicit consent. This also applies to people who are merely mentioned. Without this consent, you may not upload the content concerned; remove or anonymise it beforehand so that the person cannot be identified, not even from the context. Without a specific legal permission, you may also not upload identifiable information about criminal convictions, investigations or alleged offences of other people.

Before recording or uploading, provide the people concerned with our Notice for other people in recordings, including the Privacy Policy linked there. You must document required consents and be able to prove them upon legitimate request. Your own consent does not replace the consent of other people. Our statutory information obligations remain unaffected.

The user who records or uploads gets access to the recording and the results. Further recipients are set out in section 7; storage and deletion follow section 8. Data subjects can exercise their rights under section 10 at info@ailloy.io, even without a sahmo account. We may ask for information needed to locate their data and take the rights of other people into account.

7. Recipients of data and transfers to third countries

7.1 Recipients of your data

We use selected external service providers to provide and develop the sahmo service. Where they process personal data on our behalf, this takes place on the basis of corresponding agreements pursuant to Art. 28 GDPR.

These include in particular:

The processors used for transcription and content analysis process the transmitted content exclusively to provide the agreed services and not to train AI models.

7.2 Transfers to third countries

Our primary storage of app user data takes place within the European Union. To carry out the AI-supported functions, we transmit the necessary audio recordings or transcripts and content data to the service providers named above for that purpose.

Where internationally active providers are involved, personal data may also be processed outside the European Economic Area, in particular in the USA. A storage location in the EU does not automatically rule out such transfers, for example in the context of support access. Depending on the service used, this also concerns the providers for the website, sign-in, analysis and error diagnosis. For marketing attribution under section 14.2, Meta and TikTok may also process data outside the European Economic Area as independent controllers.

In these cases, transfers take place on the basis of the requirements of Art. 44 et seq. GDPR, in particular:

You can obtain information about the safeguards used for a specific transfer and a copy of the relevant safeguards through the contact details given in section 2.

7.3 Disclosure to authorities

We disclose personal data to authorities or other bodies where we are legally obliged to do so, or where this is necessary and legally permissible to assert, exercise or defend legal claims. The legal basis is in particular Art. 6 (1) (c) or (f) GDPR; for special categories of personal data, the requirements of Art. 9 GDPR must additionally be met.

8. Storage period and deletion

We store personal data only for as long as this is necessary for the respective purposes or as long as statutory retention obligations exist.

Deletion by you. You can delete an individual recording and your entire user account in the app. After confirmed deletion, access to the recording, the associated audio, the transcript and the analysis, as well as access to the account, ends permanently, including with access tokens issued earlier. If you have already uninstalled the app, you can request deletion of your account.

Retention afterwards. In the current product iteration, the server-side data of a deleted recording or a deleted account initially remains stored physically until it is removed as part of a later clean-up. We do not currently commit to a fixed period for this. Upon a deletion request, we remove the associated data.

Local data on your device. During upload and processing, the recording also remains on your device. Once completion is confirmed, the app removes the copy that is no longer needed there. After a confirmed account deletion, the app also clears paused recordings, failed recordings and pending uploads of that account.

Technical server logs. These are generally stored for up to 30 days, unless longer storage is exceptionally necessary to investigate specific security incidents.

Data for content quality assurance. Processing takes place only within the scope of your consent and for as long as it is necessary for that purpose. After withdrawal, further use for this purpose ends; personal copies are deleted unless other lawful retention is required. Where possible, pseudonymised or anonymised analyses are used for quality assurance.

Data held by service providers. When your account is deleted, we also remove the associated customer record at RevenueCat. This does not cancel a subscription concluded through the App Store or Google Play; you can only do that there yourself. Data already collected by Adjust is not deleted when your account is deleted. The retention periods of the other service providers named in section 7.1 additionally apply.

8.1 Temporary voice characteristics

The speaker embeddings created at AssemblyAI are used exclusively for speaker separation within the respective transcription job. According to AssemblyAI, they are not retained after the transcript has been delivered and are not reused for other recordings. We ourselves do not receive or store speaker embeddings, only the transcript with speaker labels.

We delete every transcription job that AssemblyAI has accepted there by means of a DELETE request as soon as it is finished, including when it failed or no speech was detected. If processing stops on our side while the job is still running at AssemblyAI, we keep checking it and delete it as soon as it is finished. If you delete a recording, we also delete the associated jobs at AssemblyAI. If a deletion request fails, we repeat it. If, after a technical fault, it remains unclear whether AssemblyAI accepted a job and AssemblyAI does not report the job to us afterwards, we do not know its identifier and cannot send a deletion request. AssemblyAI has not separately confirmed to us how long it retains speaker embeddings for failed or aborted jobs; in this respect, AssemblyAI’s deletion rules apply.

The separate storage of audio recordings, transcripts and analyses follows the remaining information in section 8.

9. Data security

We use appropriate technical and organisational measures to protect your personal data against loss, misuse, unauthorised access and unauthorised disclosure.

These include in particular:

Productive systems, logs and error reports are configured so that sensitive content such as audio content, transcripts and authentication data is not captured unnecessarily. Where we use service providers, we contractually oblige them to comply with appropriate data protection and security measures.

10. Your rights

The following rights apply regardless of where you live and also without a sahmo account. You can send requests to info@ailloy.io. We generally respond within one month; we will inform you within this period of any legally permitted extension and the reasons for it. Requests are generally free of charge; statutory exceptions remain reserved.

As a data subject, you have the following rights within the statutory requirements:

You also have the right to object at any time to the processing of your personal data on grounds relating to your particular situation, where we process it on the basis of Art. 6 (1) (e) or (f) GDPR (Art. 21 GDPR); this also applies to any profiling based on those provisions.

To exercise your rights, you can contact us at any time, for example by email to info@ailloy.io or using the contact details given in section 2.

10.1 Right to lodge a complaint with a supervisory authority

You also have the right to lodge a complaint with a data protection supervisory authority about our processing of your personal data (Art. 77 GDPR).

The competent data protection supervisory authority for Wiesbaden is:

Der Hessische Beauftragte für Datenschutz und Informationsfreiheit (HBDI)
Postfach 3163
65021 Wiesbaden, Germany
Phone: +49 611 1408-0
Email: poststelle@datenschutz.hessen.de
Website: datenschutz.hessen.de

11. No automated decision-making in individual cases

AI-supported systems are used within the sahmo service to transcribe audio recordings and analyse their content in a structured way. There is no automated decision-making within the meaning of Art. 22 GDPR that produces legal effects concerning you or similarly significantly affects you.

The analyses provided serve informational purposes and self-reflection only and do not constitute binding decisions.

Content-related analyses such as summaries, classifications and observations, including those on collections, are generated by an AI language model from the transcripts. They may contain errors; the app labels the analysis of a recording as created with AI. We calculate speaking shares, pauses, word counts and comparable metrics from the transcript and timestamps without a language model.

12. Website: hosting

This website is hosted externally. The personal data collected on this website is processed on the servers of the host. This may include in particular IP addresses, website visits, server log data, meta and communication data, and other data generated through the website.

External hosting serves the purpose of providing our website securely, quickly and efficiently. Where necessary to provide our online service, processing takes place on the basis of Art. 6 (1) (b) GDPR, and otherwise on the basis of our legitimate interest in the secure, stable and efficient provision of our online service by a professional provider pursuant to Art. 6 (1) (f) GDPR.

We use the following host:

Railway Corp
548 Market St PMB 68956
San Francisco, California 94104, USA
Privacy policy: Railway

EU representative: DP-Dock GmbH, attn. Railway Corp, Ballindamm 39, 20095 Hamburg, Germany.

Domain registration and DNS management are handled through:

IONOS SE
Elgendorfer Straße 57
56410 Montabaur
Germany
Email: info@ionos.de

13. Website: cookies, tracking and analysis

13.1 No cookies

Our website does not set cookies and does not store or read any information on your device. It does not use analysis or tracking tools and does not load any third-party content. Consent is therefore not required.

13.2 Server log files

The provider of the website automatically collects and stores information in so-called server log files, which your browser automatically transmits to us. These are in particular:

This data is not merged with other data sources. It is collected for the technical provision of the website, to ensure stability and security, and for error analysis on the basis of Art. 6 (1) (f) GDPR.

Our legitimate interest lies in the secure, stable and error-free provision of our website.

The download buttons and the QR code on our website do not lead directly to the App Store or Google Play, but through a link of the service Adjust, which then forwards you. In doing so, Adjust processes in particular your IP address, technical device and browser information and the time of the click, in order to attribute a later app install to the campaign. The legal basis is our legitimate interest in measuring the success of our marketing activities pursuant to Art. 6 (1) (f) GDPR. The QR code itself is served from our own server and does not load any third-party content. Section 14.2 describes the further processing by Adjust in the app.

13.4 Fonts

This website uses Google Fonts for a consistent display of typefaces. The Google Fonts are embedded locally on our server. No connection to Google servers is established when the fonts are loaded. As the fonts are hosted locally, no personal data is transmitted to Google in this respect.

14. App: tracking and analysis

In our mobile app we use analysis and crash reporting tools to improve the technical stability, security, performance and usability of the app. This may process in particular usage data, technical data, device information, event data, error reports and crash reports.

14.1 Sentry: crash reporting and error analysis

We use Sentry in the app and in the backend, a service for error analysis, crash reporting and application monitoring. The provider is Sentry, Inc.

Sentry helps us detect, analyse and fix technical errors, crashes, performance problems and security-relevant events. The following data may be processed in particular:

Audio content, transcripts, analyses, inputs and outputs of the AI systems, request content, signed links and access tokens are not transmitted to Sentry. Sentry receives technical error reports, but no recordings of the screen or session flow. Location information based on the IP address is suppressed.

The legal basis is Art. 6 (1) (f) GDPR. Our legitimate interest lies in the secure, stable and error-free provision of the app and in quickly detecting and fixing technical problems.

14.2 Adjust: mobile attribution and campaign measurement

We use Adjust in our app, a service of adjust GmbH, Saarbrücker Straße 37a, 10405 Berlin, Germany, to measure the success of our app install campaigns and to analyse the effectiveness of our marketing activities. Adjust also serves to detect and prevent fraudulent installs and ad interactions.

Adjust measures app installs, sessions and, at most once per install, the events “first recording submitted”, “first analysis viewed” and “account created”. The following data may be processed in particular:

Adjust does not receive an identifier of your sahmo account, an identifier of a recording, audio content, transcripts or analyses. Attribution takes place via the install.

Adjust only starts with your consent. On Android, we ask for it with a separate, optional checkbox when you first open the app; with a user account, you can give or withdraw it later in the account settings. On iOS, Adjust only starts if you allow tracking in the system prompt for App Tracking Transparency; you can revoke this permission at any time in the iOS settings. Without consent, Adjust does not start. The legal basis is your consent pursuant to Art. 6 (1) (a) GDPR and Section 25 (1) TDDDG. After withdrawal, the SDK is disabled and no further events are transmitted. The lawfulness of processing up to the withdrawal remains unaffected; data already collected is not deleted by the withdrawal.

If you have agreed, Adjust shares the install and the three events named above, together with pseudonymous device and advertising identifiers, with Meta and TikTok. This lets us measure which ads shown there lead to installs and to the use of sahmo. These identifiers may allow your device to be recognised across different apps. The content of your recordings, transcripts, analyses, your email address, purchases and amounts are not transmitted. Meta and TikTok process the data as independent controllers under their own privacy policies.

Regardless of your consent and of Adjust, the app on iOS registers once with Apple’s ad attribution (SKAdNetwork or AdAttributionKit) when it first starts. If you tapped an ad for sahmo before installing, Apple sends the ad network concerned a delayed install notification without any user or device identifier; Adjust receives a copy of this notification. The legal basis is our legitimate interest in measuring the success of our marketing activities pursuant to Art. 6 (1) (f) GDPR.

You can find further information about the processing by Adjust in Adjust’s privacy policy.

14.3 RevenueCat: subscriptions and entitlements

We use RevenueCat, a service of RevenueCat Inc., to manage in-app purchases, subscriptions and entitlements.

This processes in particular pseudonymous user or device identifiers, subscription and entitlement data, and technical transaction and purchase information. Purchase and payment are handled through the Apple App Store or Google Play.

Processing takes place on the basis of Art. 6 (1) (b) GDPR, where it is necessary to carry out the usage relationship and to provide the respective functions. For analysis functions beyond this that require consent, we obtain separate consent.

You can find further information in RevenueCat’s privacy policy.

15. Additional information for persons in the United States

Because we are established in Germany, the GDPR also applies to the processing described here. Applicable US privacy laws supplement these requirements and your rights. Our primary storage takes place in the European Union. International service providers may also process data outside the EU; details are set out in section 7.2.

Health data

Recordings, transcripts and analyses may be protected as consumer health data under US law. Categories, sources, purposes, recipients and rights are described in our separate Consumer Health Data Privacy Policy.

Voices

Speaker separation based on temporary voice characteristics is described in section 3.3. Section 8.1 applies to their retention.

Marketing measurement

Marketing measurement is off by default. Only if you agree, on Android via the optional checkbox and on iOS via the tracking prompt, do we use Adjust to measure which ads lead to installs and to the use of sahmo. For this purpose, Adjust transmits the install and the events “first recording submitted”, “first analysis viewed” and “account created”, together with pseudonymous device identifiers, to Meta and TikTok. These identifiers may allow your device to be recognised across different apps. We do not transmit the content of recordings, transcripts, analyses, your email address, purchases or amounts. We do not determine which state you are in; once you agree, measurement works the same everywhere. Details are set out in section 14.2.

You can turn off marketing measurement at any time: on Android with a user account in the app’s account settings, on iOS in the iOS settings under Tracking. This stops future sharing and also serves as your opt-out where this sharing is considered a sale or sharing for targeted advertising under applicable US law. Beyond this, we do not sell your personal data or share it for targeted advertising.

Your rights and appeals

The rights under section 10 also apply to you and can be exercised without a sahmo account. The rights under applicable US privacy law apply in addition. We generally respond to requests within one month and free of charge; we will inform you in good time of any legally permitted extension and the reasons for it.

You can appeal a full or partial refusal of your request by replying with the subject “Appeal”. We will review the decision again and respond within one month, stating our reasons. If the refusal is upheld, we will tell you how to complain to the competent authority, in particular the Attorney General of your state. Complaints are accepted by, among others:

16. Changes to this Privacy Policy

We may adjust this Privacy Policy from time to time to reflect changed legal, technical or organisational conditions. The current version is available in the app and on our website. In the case of material changes, we will inform you in an appropriate manner.