01Scope & contact
Sections 1–16 of this policy apply to Recall's iPhone features, including iPhone Duo layout adaptations, as well as iPad and companion Apple Watch features. See Section 17 for the separate Mac disclosure. Features may vary by platform, version, and device. Privacy contact: recallwatch.support@gmail.com.
02Starting recording & on-device records
The app captures audio only after you start listening. Inform participants and obtain any necessary consent. Audio is used for transcription. Text, people labels, summaries, to-dos, notes, journal entries, self-assessments, and content you enter are stored on your device. AI output may be incorrect; check important details.
03Temporary audio & retention
On mobile devices, after transcription succeeds and the text is saved, the app removes the corresponding temporary audio. Watch audio is removed after it is successfully transferred to the paired phone. Failed or pending items may be stored temporarily for retries. After 7 days, they are removed the next time cleanup runs on the device; untranscribed content may not be recoverable. Cleanup cannot be guaranteed to occur on time while the device is off or the app has been force-quit. Text records are retained until you delete them; Recently Deleted and backups may contain additional copies. Deleting a file is not the same as forensic-grade erasure. See Section 17 for how Mac audio held in memory is handled.
04On-device processing & model downloads
When you choose on-device speech or text models, the corresponding processing runs on your device. A local processing failure does not automatically switch to the cloud. Models may require an internet download before first use, and the download service may receive connection information such as your IP address. Apple's system-provided on-device models depend on device and operating system support.
05Optional cloud speech processing
After you choose cloud transcription, configure a speech API key, and allow uploads, audio is sent to the configured official Alibaba Cloud HTTPS service for transcription. If you select and allow Apple's online speech recognition, Apple speech services may receive recordings. Third-party processing locations, retention periods, and usage policies are determined by those services, which may charge separately.
06Optional text analysis & online verification
After you select API mode and allow analysis, conversation text, questions, relevant context, or journal content is sent to your chosen service as needed for summaries, translation, questions and answers, corrections, reviews, or profiles. In DeepSeek mode, Points to Explore can also select technical terms and public factual claims, submit the relevant words or passages to the DeepSeek search API, and organize the sources. This filtering is intended to exclude private names, health information, travel plans, and similar content, but AI filtering is not an absolute guarantee. On-device text analysis does not automatically switch to an API service. Speech transcription and text analysis settings are managed separately.
07Health, mood & profiles
With your permission, the app can read heart rate and heart rate variability (SDNN), and use self-assessments, notes, and saved statistics to support reviews. If you allow API analysis that includes health signals, the required statistics and summaries of your personal reference data are sent. This does not automatically give features such as session questions and answers access to health samples. Health and personality observations are for personal reflection only, not diagnosis, treatment, or formal personality assessment. You can manage health permissions in system settings.
08Location
Location recording requires you to enable it and grant permission. Location is collected while listening. The feature that resolves and saves place names may send coordinates to Apple Maps services; it does not send conversation or health content to those services. Location observations are processed on your device. Standard session questions and answers using your selected model do not include location samples. You can disable location access, location recording, or place-name resolution.
09People & API keys
Speaker differentiation and, if you choose to remember a voice, voice characteristics are processed and stored on your device. You can manage or delete them. People labels may be synced or analyzed with text records. API keys are stored in the device's Keychain and are not synced as part of recordings or journal content. Do not include keys in feedback.
10Sync, exports & backups
After you connect an encrypted sync folder, conversation text, people labels, summaries, locations, translations, and daily reviews, including health statistics and analysis already generated, can sync to a location you choose, such as iCloud Drive. Sync does not include API keys, models, voiceprints, or original audio files. The app also offers backups and exports with different scopes, which may include full text records, topics, journals, locations, and more. Refer to the scope listed on the export screen; do not assume a full backup contains only journal entries. Keep passwords for encrypted sync and encrypted backups safe. After export, you decide where files are saved, who they are shared with, and how copies are deleted.
11Membership & purchases
Membership products and transactions are retrieved, purchased, and verified through Apple StoreKit. The app reads the relevant transaction and entitlement status to provide membership features. Trial start dates and status are stored in the local Keychain. Apple processes payment information; do not provide payment card details in feedback. Third-party API fees are separate from Apple membership purchases.
12Feedback & third-party services
The email address, issue description, and attachments you voluntarily send to the support inbox are used by the developer and email service to handle your feedback. If you leave comments on Xiaohongshu or submit TestFlight feedback, those platforms process the content under their own rules. Provide only the information needed to resolve the issue. The developer cannot directly delete copies stored on your devices or in third-party services you choose.
13Your choices & deletion
You can stop recording, switch analysis methods, turn off optional features, revoke system permissions, delete records and locally remembered people, or disconnect sync. Also check Recently Deleted, other devices, backups, and exported files. Deleting local content does not guarantee that existing copies held by sync services, email services, or third-party APIs are deleted at the same time. You can send requests to access or delete feedback emails to the support address.
14Children's & participants' information
Recordings and notes may involve other people or children. Record, process, or share information only with appropriate authorization and consent. Avoid sending unnecessary private content to cloud services or public channels.
15Policy updates
Updated November 13, 2026. This policy describes data processing based on the current version's features. Versions and devices may differ; read it alongside your actual settings.
16This website
This support website is hosted on Cloudflare. When you visit, the hosting service processes network request information, such as IP addresses, request times, and browser information, to serve the pages and maintain security. Its handling of this information is governed by its service policies. The pages contain no added advertising, third-party analytics scripts, online forms, or external fonts. If you follow an email or Xiaohongshu link, the relevant service processes information under its own rules.
17Mac version
The Mac version is a separately installed native app, and its features and purchase entitlements may differ from the mobile app. Microphone audio is captured only after you start listening. The recording buffer is held only in memory; no audio recording files are written. Untranscribed audio cannot be recovered after you quit the app, and listening may also stop if the processing queue becomes too full. The mobile audio retention rule of 7 days described above therefore does not apply. Depending on your settings, text analysis can use local models, available Apple on-device models, or an API you configure yourself. If you choose an API, relevant text and context are sent to the selected service, which may charge separately. Model downloads require an internet connection. You can manage saved text records, actions, and reviews on your device. If you actively set up encrypted folder sync, relevant content such as text is synced according to the features available; API keys, models, voice characteristics, and original audio are not synced. Cross-device sync is subject to system scheduling and is not guaranteed to be immediate or to run continuously in the background. Manage permissions, sync, deletion, and third-party copies according to the features you actually use. Do not assume that mobile-only features are available on Mac.