Hyperdrive
Features Pricing Changelog
Legal

Privacy Policy

Effective Date: February 21, 2026 · Last Updated: March 10, 2026 · Version 2.5

Effective Date: February 21, 2026
Last Updated: March 10, 2026
Product: Hyperdrive, a native macOS email client
Developer: Hyperdrive One LLC
Policy Version: 2.5


This Privacy Policy explains in detail how Hyperdrive One LLC ("we," "us," "our," or "Hyperdrive") collects, uses, stores, shares, discloses, retains, and protects information when you ("you," "your," or "user") use the Hyperdrive macOS application ("Hyperdrive" or "the app"), related server-side services operated by Hyperdrive, and associated websites, APIs, and web-based features (collectively, "the Service").

By downloading, installing, accessing, or using Hyperdrive, you acknowledge that you have read, understood, and agree to the practices described in this Privacy Policy. If you do not agree with any part of this Privacy Policy, you must discontinue use of the Service.

This Privacy Policy should be read in conjunction with our Terms of Service, which govern your use of the Service.


Table of Contents

  1. Scope and Applicability
  2. Definitions
  3. Summary of Data Collection Practices
  4. Information You Provide Directly
  5. Information Hyperdrive Collects Automatically — Local Storage
  6. Information Collected and Stored on Hyperdrive Servers
  7. How Hyperdrive Uses Information
  8. Legal Bases for Processing
  9. Third-Party Services and Data Sharing
  10. Artificial Intelligence and Machine Learning Features
  11. Email Tracking and Recipient Data
  12. Automated Decision-Making and Profiling
  13. Local-Only Processing
  14. Hardware Identifiers, Device Fingerprinting, and Licensing
  15. Notifications
  16. Data Retention
  17. Data Deletion, Export, and Account Controls
  18. Security Measures
  19. International Data Transfers
  20. Your Rights Under the General Data Protection Regulation (GDPR)
  21. Your Rights Under the California Consumer Privacy Act (CCPA/CPRA)
  22. Do Not Track Signals
  23. Children's Privacy
  24. Information We Do Not Collect
  25. Your Responsibilities
  26. Server-Side Deferred Actions and Expanded Trust Boundary
  27. Changes to This Privacy Policy
  28. Contact Information

1. Scope and Applicability

This Privacy Policy applies to:

  • The Hyperdrive desktop and mobile applications, including the macOS and iOS apps, and all features, modules, and functionality delivered through the application binaries.
  • Server-side services operated by Hyperdrive that support the application, hosted at app.hyperdriveone.com.
  • Web-based scheduling pages served from app.hyperdriveone.com that allow third parties to book meetings with Hyperdrive users.
  • Any features within Hyperdrive that connect to or exchange data with third-party services, including but not limited to Google (Gmail, Calendar, People/Contacts, and Meet APIs), Zoom, Slack, OpenAI, HuggingFace, Gravatar, Unsplash, Stripe, and Cloudflare.

This Privacy Policy does not apply to:

  • Third-party products, services, websites, or applications that you connect to or access through Hyperdrive. Those third parties maintain their own privacy policies and practices, and Hyperdrive is not responsible for their data handling.
  • Email content or communications created, sent, or received by you through third-party email providers such as Google Gmail. The handling of such content by those providers is governed by their respective privacy policies.
  • Any custom AI endpoint you choose to configure under the Custom AI tier. If you provide your own API key and optional endpoint URL, data transmitted to that endpoint is governed by the privacy practices of the endpoint operator, and you assume full responsibility for evaluating its privacy and security posture.

2. Definitions

For purposes of this Privacy Policy, the following terms have the meanings set forth below:

  • Account means any email account, messaging workspace, video conferencing service, or other service credential you connect to Hyperdrive, including but not limited to a Google account (for Gmail, Google Calendar, and Google Contacts access), a Slack workspace (for Slack messaging integration), and a Zoom account (for video conferencing and scheduling link meeting creation).
  • Content means the text, headers, subject lines, body content, attachments, metadata, timestamps, labels, sender and recipient addresses, unsubscribe headers, and other information contained in or derived from emails, Slack messages, calendar events, drafts, scheduled messages, snooze and reminder configurations, or contact relationship data that are processed by Hyperdrive.
  • Device Identifier means a persistent or semi-persistent identifier associated with your hardware, including but not limited to the macOS platform UUID derived from IOKit (IOPlatformExpertDevice / kIOPlatformUUIDKey), used for licensing activation and device limit enforcement.
  • Local Storage means data stored on your Mac, including data stored in the macOS Keychain, SQLite databases and configuration files within ~/Library/Application Support/Hyperdrive/, UserDefaults preferences at ~/Library/Preferences/com.hyperdrive.app.plist, application cache files at ~/Library/Caches/Hyperdrive/, and machine learning model caches within ~/Library/Caches/ managed by third-party model loading libraries.
  • Server-Side Storage means data stored on Hyperdrive infrastructure, which consists of Cloudflare Workers for compute, Cloudflare D1 (managed SQLite at edge) for structured persistent storage, and Cloudflare R2 (S3-compatible object storage) for binary data such as email attachments associated with scheduled or deferred sends. If Hyperdrive migrates to different infrastructure providers, this Privacy Policy will be updated to reflect the change, and users will be notified in accordance with Section 23 (Changes to This Privacy Policy).
  • Personal Information or Personal Data means any information that identifies, relates to, describes, is reasonably capable of being associated with, or could reasonably be linked, directly or indirectly, to an identified or identifiable natural person. This includes, without limitation, email addresses, names, IP addresses, device identifiers, OAuth tokens, email content, Usage Data, and inferences drawn from any of the foregoing to create a profile about a consumer reflecting the consumer’s preferences, characteristics, behavior, or attitudes.
  • Usage Data means information collected about how you interact with Hyperdrive, including but not limited to: AI feature usage logs (feature name, model identifier, token counts, estimated cost, and timestamp, stored server-side on a per-account basis); email classification correction history (sender, subject, original category, corrected category, and timestamp, stored locally in ~/Library/Application Support/Hyperdrive/training_corrections.jsonl and never transmitted to any server); recent search queries (stored locally via UserDefaults); licensing validation metadata (license key, machine identifier, machine name (which may contain your name, e.g., “John’s MacBook Pro”), platform, application version, and IP address, transmitted to the Hyperdrive server at activation and approximately every 24 hours thereafter); and in-session activity statistics (such as counts of emails archived, snoozed, replied to, or trashed, stored in memory only and not persisted). Usage Data does not include the substantive content of emails, messages, or calendar events, which is defined separately as Content.
  • Tracking means email open tracking and link click tracking performed by injecting a one-pixel transparent image (tracking pixel) and wrapping outbound hyperlinks through a redirect endpoint in outgoing emails sent by the user. Tracking collects data about email recipients, including Recipient Data as defined below. Recipients are third parties who have not agreed to this Privacy Policy. See Section 11 for full details on Tracking data collection, recipient notice, and your responsibilities as the sender.
  • Recipient Data means information collected about recipients of emails sent by Hyperdrive users through the Tracking feature. Recipient Data includes the recipient’s email address (as provided by the sender), IP address (truncated to the network prefix: /24 for IPv4, /48 for IPv6), user-agent string of the recipient’s email client or browser (which may reveal device type, operating system, and software version), timestamps of open and click events, and destination URLs of clicked links. Recipient Data is collected without the recipient’s explicit knowledge or consent. Recipients are not parties to this Privacy Policy and do not have a direct relationship with Hyperdrive. The Hyperdrive user who sends a tracked email is solely responsible for complying with all applicable laws governing recipient tracking, notice, and consent. See Section 11 for full details.
  • Processing means any operation performed on Personal Data, whether or not by automated means, including collection, recording, organization, structuring, storage, adaptation, alteration, retrieval, consultation, use, disclosure by transmission, dissemination, alignment, combination, restriction, erasure, or destruction.
  • Proxy or Server Proxy means the Hyperdrive server infrastructure that acts as an intermediary between the Hyperdrive client application and a Third-Party Service Provider (such as OpenAI, Unsplash, Gravatar, Google Calendar, Zoom, or Slack). The Proxy injects API credentials, meters usage, sanitizes and validates requests, and routes requests without exposing third-party API keys to the client application. In addition to transient request routing, the Proxy may store encrypted OAuth refresh tokens on behalf of the user (for deferred actions such as scheduled sends, snoozes, and reminders), persist email attachments in object storage for deferred delivery, cache third-party API responses (such as contact enrichment data), and log usage metadata (such as AI feature usage, token counts, and estimated cost) on a per-account basis. See Section 6 for full details on server-side data collection.
  • Third-Party Service Provider means any external entity that receives, processes, or stores Personal Information on behalf of or in connection with Hyperdrive’s operation. As of the date of this Privacy Policy, Hyperdrive’s Third-Party Service Providers include: Google LLC (Gmail, Calendar, People/Contacts, and Meet APIs; receives email content, calendar data, contact data, and OAuth credentials when you connect a Google account); Slack Technologies, LLC (workspace messaging integration; receives OAuth credentials and accesses channel and direct message history when you connect a Slack workspace); Zoom Video Communications, Inc. (video conferencing; receives OAuth credentials and meeting metadata when you connect a Zoom account for scheduling link meetings); Stripe, Inc. (payment processing; receives email address, machine identifier, machine name, and payment information); OpenAI, LLC (AI-assisted features such as draft generation and email summarization; receives email and Slack content on an opt-in basis through the Server Proxy); Resend, Inc. (transactional email delivery; receives user email addresses for license confirmations, recovery codes, and support correspondence); Cloudflare, Inc. (infrastructure provider; all server-side data is processed and stored on Cloudflare Workers, D1, and R2 infrastructure); Automattic, Inc. (Gravatar) (contact enrichment; receives SHA-256 and MD5 cryptographic hashes of email addresses, not raw addresses); Unsplash / Getty Images (daily photo feature; does not receive Personal Information); and Hugging Face, Inc. (machine learning model distribution; the client downloads model weights via standard HTTPS without transmitting Personal Information). Each Third-Party Service Provider operates under its own privacy policy and data processing terms. Hyperdrive does not sell Personal Information to any Third-Party Service Provider. See Section 13 for full details on third-party data sharing.
  • Vector or Inbox Vector means a categorized view of the inbox (such as Priority, Team, Newsletters, Transactions, Shipping, Calendar, or Backlog) that organizes emails based on a multi-layer classification pipeline including user overrides, deterministic rules, weighted heuristic scoring, on-device machine learning models, and—as a platform fallback for devices that cannot run on-device models—cloud-based AI classification.
  • On-Device ML Model means a machine learning model that runs entirely on your Mac without transmitting data to any external server. Hyperdrive uses two on-device models for email classification: a bundled text classifier and a separately downloaded language model. The downloaded model requires Apple Silicon and is not available on Intel-based Macs.
  • Shared Classification Intelligence means an opt-in feature that aggregates anonymized sender category votes across Hyperdrive users using differential privacy protections, cryptographically hashed sender tokens with periodically rotating salts, and a minimum vote threshold per sender before results are returned.

3. Summary of Data Collection Practices

Hyperdrive is an email client that necessarily processes sensitive personal communications data. The scope of data collection is broad because providing full-featured email client functionality, AI-assisted productivity features, licensing enforcement, and cross-service integrations requires access to and processing of personal information.

The following is a high-level summary of the categories of information Hyperdrive collects and processes. Each category is described in full detail in subsequent sections of this Privacy Policy.

Category Stored Locally Stored on Server Shared with Third Parties
Email content, metadata, and thread data Yes Yes — scheduled messages stored encrypted until delivery (7-day retention); subjects and recipient addresses in tracking pixels (90-day retention); thread IDs for snoozes, reminders, and classification sync; sender domain classifications synced across devices Yes (Google, OpenAI via proxy)
Slack messages and metadata Yes No (transient only — passed through Server Proxy for AI summarization, not persisted) Yes (Slack, OpenAI via proxy for channel summarization)
OAuth tokens and credentials Yes (Keychain) Yes — Gmail and Zoom refresh tokens stored encrypted for scheduling, deferred sends, snoozes, and reminders No
Licensing and trial data Yes Yes No
Email tracking data (pixels, opens, clicks) No Yes No
Contact enrichment data No Yes (cached 30 days) Yes (Gravatar)
AI feature inputs and outputs Yes (on-device model weights, classification results, summaries cached locally) No (content not logged; cloud AI usage metered) Cloud features only (OpenAI via proxy for compose, rewrite, summarize, sender classification)
Calendar events and attendee data Yes Partially (scheduling bookings) Yes (Google Calendar)
Google Contacts data (names, emails, phone numbers, organizations) Yes (SQLite cache) No Yes (Google People API)
macOS Contacts (system address book) Yes (SQLite cache) No No
User preferences and settings Yes No No
Device identifiers and hardware UUIDs No Yes (licensing) No
IP addresses No Yes (tracking, licensing) Incidentally (Cloudflare, third-party APIs)
Search queries Yes (recent searches) No Yes (if AI search used, via OpenAI proxy)
Sender classification data Yes (on-device ML models, local SQLite cache) Yes — per-account sender domain classifications and per-thread overrides synced for cross-device consistency; opt-in differential-privacy-protected sender category votes if Shared Intelligence enabled Yes (unknown sender emails and sample subjects sent to OpenAI via proxy for batch classification; on-device models used first when available)
Scheduling link and booking data Yes Yes (guest data encrypted at rest) Yes (Google Calendar and Google Meet for event creation; Zoom for meeting creation if configured)

4. Information You Provide Directly

4.1 Account Connection Information

When you connect email, calendar, messaging, or conferencing services to Hyperdrive, you provide authorization for Hyperdrive to access those services on your behalf. As part of this process, Hyperdrive receives and stores the following information locally on your device:

  • For Gmail: your account email address, OAuth 2.0 access tokens (short-lived credentials that grant API access), OAuth 2.0 refresh tokens (long-lived credentials that allow Hyperdrive to obtain new access tokens without requiring you to re-authorize), and token expiration timestamps.
  • For Slack: workspace identifiers, user identifiers, a non-expiring user token, and team name, if you connect a Slack workspace.
  • For Zoom: access tokens, refresh tokens, token expiration timestamps, and your Zoom email address, if you connect a Zoom account for scheduling link meetings.
  • If you configure a custom AI provider: your third-party API key, stored in the macOS Keychain.

In addition, Gmail and Zoom refresh tokens are transmitted to Hyperdrive's server (app.hyperdriveone.com) to enable server-side features that operate when your Mac is offline or asleep. Specifically:

  • Gmail refresh tokens are sent to the scheduling server (for creating Google Calendar events when guests book meetings via your scheduling links) and to the deferred actions server (for server-side send-later delivery, snooze restoration, and reminder firing).
  • Zoom refresh tokens are sent to the scheduling server (for creating Zoom meetings when guests book meetings via your scheduling links).

All server-side tokens are encrypted at rest using AES-256-GCM with keys derived from per-user random salts and environment-specific encryption secrets. See Section 26 for details on server-side deferred actions.

The specific OAuth authorization scopes requested by Hyperdrive are detailed in Section 9.

4.2 Email You Write, Draft, and Schedule

When you compose, draft, reply to, forward, or schedule email within Hyperdrive, the application stores:

  • Draft content, including the full message body in plain text format, and all metadata (recipients, subject, CC, BCC fields, attachments, and threading references).
  • Scheduled messages, including the full composed email content, the scheduled send date and time, and the originating account identifier.
  • Email signatures you create, stored per-account in HTML, plain text, and RTFD formats.
  • Text snippet templates you create for quick insertion, including optional text expansion shortcuts (short trigger strings such as --intro that auto-expand when typed in the composer).

Scheduled messages may also be stored server-side for reliable send-later delivery even when your Mac is offline, including the full message content and any file attachments. See Section 26 for details on server-side deferred actions.

Snippet sync: Text snippets are synced to Hyperdrive’s server at app.hyperdriveone.com for cross-device access. Snippet names and text expansion shortcuts are stored in plaintext; snippet body content is encrypted at rest using AES-256-GCM before storage. Soft-deleted snippets (marked with a deleted flag) are purged from the server after 30 days.

Signature sync: Email signatures (HTML and plain-text body representations) are synced encrypted to Hyperdrive’s server at app.hyperdriveone.com for cross-device access. Both the HTML body and plain-text body are encrypted at rest using AES-256-GCM before storage. Platform-specific rich-text data (RTFD) remains device-local only and is not transmitted to the server.

Draft sync: Drafts are synced to Gmail via the Gmail Draft API for cross-device continuity. Draft content transits through and is stored by Google in accordance with Google’s privacy policy. Local draft metadata (including the Gmail draft identifier) is stored in the device-local SQLite database.

Retention and deletion: Draft content is stored locally on your device and is automatically deleted when the draft is sent. Scheduled messages and their attachments are deleted from the server immediately after successful delivery. Scheduled messages that fail or are cancelled are purged within 30 days. Synced snippets are retained on the server until deleted by you or until account disconnection; soft-deleted snippets are purged after 30 days. Synced signatures are retained on the server until replaced by you or until account disconnection.

4.3 Scheduling Link Information and Bookings

If you use scheduling links and booking features, Hyperdrive stores:

  • Scheduling link configuration, including: title, URL slug, duration in minutes, buffer time between meetings, availability windows (days and hours as JSON), selected calendar identifier, timezone, conference provider (Google Meet or Zoom), and active/inactive status.
  • Booking details when a guest books a meeting with you, including: guest name, guest email address, booking start and end times, and the identifier of the Google Calendar event created. Guest name and email are encrypted at rest on the server.
  • Your Google Calendar OAuth access and refresh tokens and your Zoom OAuth access and refresh tokens, encrypted server-side using AES-256-GCM with keys derived via PBKDF2 (SHA-256, 100,000 iterations) from per-user random salts and the SCHEDULING_ENCRYPTION_KEY environment variable. These tokens enable the server to create calendar events and video conference meetings on your behalf when a guest books a meeting.

Retention and deletion: Booking details (including encrypted guest name and email) are automatically purged 30 days after the meeting end time. Scheduling link configuration persists until you delete the link or delete your account. Expired OAuth access tokens are automatically cleared on each server cleanup cycle; refresh tokens persist (encrypted) for ongoing calendar and conferencing access until you disconnect the integration or delete your account, at which point all scheduling data (links, bookings, and tokens) is permanently deleted.

4.4 Inbox Configuration

When you customize your inbox, Hyperdrive stores:

  • Vector configuration files, including custom inbox vector definitions and filtering rules, stored as JSON files locally at ~/Library/Application Support/Hyperdrive/ (one file per connected account).
  • Thread vector overrides, sender-level overrides, and domain-level overrides, where you manually re-classify a thread, sender, or domain into a different vector.
  • Blocked sender lists, including the blocked email address and the timestamp of the block action.
  • Muted thread identifiers for conversations you have muted.
  • On-device machine learning model weights: bundled text classifiers and, if downloaded, a language model stored in a system-managed cache directory within ~/Library/Caches/. The downloaded model requires Apple Silicon.

In addition, vector configurations, sender classifications, and thread vector overrides are synced encrypted to Hyperdrive’s server at app.hyperdriveone.com for cross-device consistency. Sender classifications are stored by domain and account. Thread overrides are stored by thread identifier and account. Vector configuration is stored as one encrypted JSON blob per account. All synced classification data is deleted when you disconnect an account.

4.5 Support Communications

If you contact Hyperdrive for support via the web form, you provide:

  • Your email address (must match a licensed account).
  • A support category (bug report, feature request, billing, or general).
  • A subject line (up to 200 characters) and message body (up to 5,000 characters).

Your license identifier and license status are automatically attached to the support email sent to our team. The content of support form submissions is not stored in our database; submissions are forwarded via email to our team using Resend (our email delivery provider) and are not retained on our servers after delivery. Your IP address is temporarily recorded for rate limiting purposes and is automatically purged.

If you contact us directly at support@hyperdriveone.com, you may also share device details, screenshots, logs, or other diagnostic information as email attachments.

We use support communications solely to respond to your inquiry and to improve the Service. We do not use support communications for marketing purposes.

4.6 Website Signup and Waitlist

If you provide your email address through the Hyperdrive website — such as by joining the waitlist, signing up for early access, or requesting a download link — Hyperdrive collects and stores:

  • An HMAC-SHA256 hash of your email address, used for deduplication. This hash is irreversible and cannot be used to recover your email address.
  • An AES-256-GCM encrypted copy of your email address, used to send you download links, product updates, and announcements. The encrypted email can only be decrypted by Hyperdrive servers using environment-specific encryption keys.

Your IP address is temporarily recorded for rate limiting purposes and is automatically purged.

How this information is used:

  • Sending a waitlist confirmation email at the time of signup.
  • Sending download links when the product is available.
  • Sending product updates and announcements about the Hyperdrive Service.

Every product communication email includes an unsubscribe link. You may opt out at any time by clicking the unsubscribe link, after which no further product communications will be sent to your address. Hyperdrive does not sell, rent, or share website signup email addresses with third parties for their own marketing purposes.


5. Information Hyperdrive Collects Automatically — Local Storage

5.1 Local SQLite Database

Hyperdrive maintains a local SQLite database (hyperdrive.sqlite) as a cache and feature store. The database is located at ~/Library/Application Support/Hyperdrive/ and operates at schema version 36 as of the date of this Privacy Policy. The database uses Write-Ahead Logging (WAL mode) for concurrent access. This database stores the following categories of information:

Email data:

  • Email thread identifiers, subjects, snippets (short preview text), message counts, label associations, and thread-to-label junction records.
  • Full message bodies stored as binary large objects (BLOBs), along with sender and recipient information, dates, and attachment metadata.
  • Draft message data (BLOB), subject lines, recipients summary, timestamps, and Gmail draft identifier (gmail_draft_id) for server-side draft synchronization.
  • Scheduled message payloads for send-later functionality, including the full composed message content, scheduled send date, retry count, and creation timestamp.
  • Full-text search index content derived from email subjects, body text, sender names, and snippets, built using SQLite FTS5 with Unicode61 tokenization and diacritics removal.
  • Per-account sync state checkpoints, including sync cursor keys and last-sync timestamps, used to resume incremental synchronization with Gmail.

Contact data:

  • Contact email addresses, display names, interaction counts, and last-seen dates.
  • Sender relationship records (stored separately from contacts), including per-sender email counts sent and received, last sent and received dates, sender domain, and co-recipient frequency.
  • Blocked sender email addresses and block timestamps.
  • Records of senders where unsubscribe prompts were dismissed.
  • Google Contacts synced via the People API (google_contacts table): resource name, account identifier, display name, given name, family name, primary email, all email addresses (JSON), primary phone number, all phone numbers (JSON), company, job title, etag, and a flag indicating whether the contact is in the user’s “My Contacts” group. Synced incrementally every 15 minutes using a sync token stored in the macOS Keychain. No Google Contacts data is transmitted to Hyperdrive servers.
  • macOS Contacts (macos_contacts table): identifier, display name, given name, family name, email addresses, phone numbers, company, and job title, read from the system address book via the Contacts framework (CNContactStore). Refreshed periodically. Used only for local autocomplete merging and never transmitted to any server.

Scheduling and workflow data:

  • Snooze records, including thread identifiers, return dates, server synchronization identifiers, and label restore state.
  • Reminder records, including thread identifiers, remind dates, and message count thresholds.
  • Pending actions queue, containing email actions (such as archive, delete, label changes) awaiting synchronization with the server, with full action payloads.
  • Scheduling link definitions (slugs, durations, availability, timezone).

AI and classification data:

  • AI-generated thread summaries, stored locally.
  • Sender classifications assigned by the classification pipeline described in Section 10.2, cached locally indefinitely.
  • Classification confidence scores for each sender.
  • Thread vector overrides, sender-level overrides, domain-level overrides, and command palette usage frequency data.

Slack data (if Slack is connected):

  • Slack channel identifiers, names, and workspace associations.
  • Slack message content and metadata.
  • Slack thread visibility state (dismiss, star, mute).

Calendar data (if calendar features are used):

  • Calendar event data, dates, times, and attendee information.
  • Calendar metadata per account.

User interface and performance data:

  • Email rendering height caches (for UI performance optimization).
  • Remote image whitelist entries (per-sender and per-domain) and “view original” formatting whitelist entries, stored in separate tables.
  • Email signatures per account (HTML, plain text, and archived rich text data).
  • Text snippet templates with text expansion shortcuts.

Additional local files (outside the SQLite database but within Application Support):

  • Email classification correction history (training_corrections.jsonl), stored as newline-delimited JSON. Each record contains a timestamp, email subject, sender email address, sender domain, original classification category, corrected category, correction scope, and thread identifier. This file is append-only, stored locally, and is never transmitted to any server.
  • Per-account vector configuration files (one JSON file per connected account, named with a sanitized form of the account identifier), containing custom inbox vector definitions and filtering rules.

5.2 Credential Storage (macOS Keychain)

Hyperdrive stores OAuth tokens and related sensitive credentials using the macOS system Keychain, backed by Apple's Security framework (Security.framework). All Keychain items are stored with the accessibility level kSecAttrAccessibleWhenUnlockedThisDeviceOnly, which means:

  • Tokens are only accessible when your Mac is unlocked.
  • Tokens are never synchronized to iCloud Keychain.
  • Tokens are bound to the specific physical device on which they are stored.

The following credentials may be stored in the Keychain:

Credential Type Keychain Service Key Sensitivity
Gmail OAuth access token com.hyperdrive.gmail.{account_id} Critical
Gmail OAuth refresh token com.hyperdrive.gmail.{account_id} Critical
Gmail token expiration com.hyperdrive.gmail.{account_id} Low
Gmail account email address com.hyperdrive.gmail.{account_id} Personal
Slack user token, workspace identifier, and workspace name com.hyperdrive.slack.{workspace_id} Critical
Custom AI API key (if configured) com.hyperdrive.ai-api-key Critical
Zoom OAuth access and refresh tokens, token expiration, and Zoom email address (if Zoom is connected) com.hyperdrive.zoom.{account_id} Critical
License key, license identifier, activation identifier, activation secret, lease token, lease signature, and clock-drift guard timestamp com.hyperdrive.license Critical
Trial validation timestamp, days remaining, clock-drift guard timestamp, trial email address, and cumulative clock drift com.hyperdrive.trial Personal
Per-account Bearer tokens for server-side deferred actions (snooze, reminders, scheduled send) com.hyperdrive.deferred Critical
Scheduling API Bearer token (if scheduling links are configured) com.hyperdrive.scheduling.{account_id} Critical

5.3 UserDefaults Preferences

Hyperdrive stores application preferences using macOS UserDefaults, persisted at ~/Library/Preferences/com.hyperdrive.app.plist. This may include:

  • Privacy settings: Remote image blocking preference (default: enabled), always-show-original-colors preference, email tracking pixel toggle (default: enabled), Shared Classification Intelligence consent acknowledgment and opt-in toggle.
  • Notification settings: Toggles for Gmail notifications, Slack notifications, reminder notifications, and notification sound.
  • Search history: Up to 10 recent search queries.
  • AI configuration: AI consent acknowledgment flag, AI enabled toggle, AI tier (basic, pro, or custom), AI provider, heavy model name, light model name, custom endpoint URL, and ML classification mode (primary or disabled).
  • Theme and typography preferences: Appearance mode, selected theme, application font family, application font size, application letter spacing, email font family, email font size, email line height, email letter spacing, and legacy typography scale (retained for migration).
  • Email behavior: Undo send delay duration, auto-advance after archive preference, page scroll percentage, and 24-hour time format preference.
  • Account state: List of connected Gmail account identifiers, active account identifier, combined inbox mode toggle, and Slack workspace metadata (JSON).
  • Slack preferences: Opted-in Slack channel identifiers per workspace.
  • Custom key bindings: Custom keyboard shortcut definitions stored as JSON.
  • Calendar preferences: Calendar visibility overrides, calendar color overrides, account color overrides, hidden calendar identifiers, and warm domain threshold.
  • Muted conversations: List of muted thread identifiers.
  • Operational state: Per-account vector configuration sync timestamps, per-account deferred sync and classification sync timestamps, per-account sent mail backfill completion flags, CoreML GPU crash guard recovery state, trial initialization sentinel, and first-launch UI hint flags. These values are used for internal synchronization and crash recovery and do not contain email content or personal communications data.

UserDefaults data is stored in a standard macOS preference file and is not encrypted beyond macOS filesystem-level protections.

5.4 Remote Image Behavior

By default, Hyperdrive blocks all remote images in emails using a Content Security Policy (CSP) that restricts image sources to data: URIs and cid: (Content-ID) inline images only. This default protects you from tracking pixels and remote image-based surveillance embedded in incoming emails.

If you choose to whitelist a specific sender (by email address) or an entire domain for remote image loading, the application relaxes the CSP to permit loading images over HTTP and HTTPS for those specific emails. A domain-level whitelist entry applies to all senders at that domain. When remote images are loaded, your device connects directly to the third-party server hosting those images (Hyperdrive does not proxy remote image requests), and the following information may be revealed to that server:

  • That you opened the email (and approximately when).
  • Your IP address and approximate geographic location.
  • Your device type, operating system version, and rendering engine characteristics (via HTTP headers).
  • Any unique identifiers embedded in the image URL by the sender.

Whitelist entries are stored locally in the SQLite database, with separate entries for sender-level and domain-level whitelisting. A separate whitelist is maintained for the “view original” email formatting feature, which permits loading the email’s original HTML rendering (including remote resources) for whitelisted senders or domains.

5.5 Unsplash Photo Cache

Hyperdrive caches a single daily photo from Unsplash for its inbox-zero celebration feature. The photo is fetched from the Hyperdrive server (which caches a daily photo selected at midnight UTC) and stored locally in two files within ~/Library/Caches/Hyperdrive/inbox-zero/:

  • daily-photo.json — Photo metadata including the Unsplash photo identifier, photographer name, photographer profile URL, photo page URL, download tracking endpoint, image URL, local filename, and the date the photo was fetched.
  • {photo-id}.jpg — The full-resolution JPEG image file downloaded from Unsplash.

The cache is refreshed when a new daily photo is available, which occurs at minimum once every 24 hours and additionally on application launch and system wake. When a new photo is fetched, any previously cached image files are automatically deleted. This cache does not contain personal information.


6. Information Collected and Stored on Hyperdrive Servers

Hyperdrive uses server-side infrastructure for specific features that require persistent state beyond your local device or that enable functionality when your device is offline. The server infrastructure consists of Cloudflare Workers (for compute), Cloudflare D1 (for managed SQLite database storage at the edge), and Cloudflare R2 (for object storage of file attachments on scheduled messages).

All server-side functionality is served by a single Cloudflare Worker at app.hyperdriveone.com. This Worker handles:

  • Application features: AI proxy, email tracking, contact enrichment, scheduling, deferred actions (snoozes, reminders, scheduled messages), classification sync, Unsplash proxy, and Slack OAuth proxy.
  • Licensing and billing: License activation, validation, trials, device management, Stripe checkout and webhook processing, and license recovery.
  • Account management: Self-service account deletion and application update feed.

6.1 Email Tracking Data

If you use email tracking features, Hyperdrive stores the following data on Hyperdrive servers:

Pixel registration data (stored when you send a tracked email):

  • Unique pixel identifier (UUID).
  • Your account identifier (your email address).
  • Recipient email address.
  • Email subject line.
  • Thread identifier.
  • Registration timestamp.

Open event data (stored when a recipient opens a tracked email):

  • Pixel identifier (linking back to the original email).
  • Open timestamp.
  • User-agent string of the recipient's email client or browser.
  • IP address of the recipient (truncated to /24 for IPv4 or /48 for IPv6 before storage; obtained from the cf-connecting-ip header provided by Cloudflare).

Click event data (stored when a recipient clicks a tracked link):

  • Link identifier (UUID).
  • Pixel identifier (linking to the email containing the link).
  • Destination URL.
  • Click timestamp.
  • User-agent string.
  • IP address (truncated to /24 for IPv4 or /48 for IPv6 before storage).

Retention: All email tracking data — pixel registration records, open events, and click events — is automatically deleted after 90 days via an automated retention cleanup process. See Section 16 for full retention details.

Encryption at rest: When a server-side encryption key is available, recipient email addresses are encrypted using AES-256-GCM before storage (in the recipient_encrypted column). In this case, the plaintext recipient column is set to null. When the encryption key is not configured, recipient email addresses are stored in plaintext. Encrypted recipient data is decrypted server-side when you query your tracking results via the authenticated API.

Data minimization note: Email subject lines are stored in the server-side pixels table. This data is not strictly necessary for open detection (which requires only the pixel identifier) but is stored to provide richer reporting to the sender.

6.2 Contact Enrichment Cache

If contact enrichment is enabled and a contact lookup is performed through Gravatar, the following data is cached server-side:

  • Contact email address (used as the lookup key).
  • Name, company, title, and location returned by Gravatar.
  • Photo URL.
  • Social profile data, including LinkedIn profile URLs extracted from Gravatar's verified accounts.
  • Data source identifier (e.g., "gravatar" or "gravatar-profile").
  • Timestamp when the data was fetched.
  • Expiration timestamp (set to 30 days after fetch).

Contact PII (name, company, title, location, social profile data, and photo URL) is encrypted at rest into a single encrypted blob using AES-256-GCM before storage. The individual plaintext columns are retained for backward compatibility with pre-encryption records but are set to null for new entries. This cache is designed to expire and be refreshed after 30 days.

6.3 Scheduling System Data

For scheduling features, Hyperdrive stores the following server-side:

  • Scheduling user records: User identifier (typically your email address), encrypted Google Calendar OAuth refresh token, encrypted access token, access token expiration timestamp, hashed API authentication token, and per-account encryption salt. If Zoom is connected: encrypted Zoom OAuth refresh token, encrypted Zoom access token, and Zoom access token expiration timestamp.
  • Scheduling link records: URL slug, user identifier, title, duration in minutes, availability windows (JSON), calendar identifier, timezone, active/inactive status, buffer time between bookings (in minutes), and conference provider setting (google_meet, zoom, or none).
  • Booking records: Booking identifier, scheduling link identifier, booking start and end times, and Google Calendar event identifier. Guest name and guest email address are encrypted at rest into a single guest_data_encrypted blob; the plaintext columns are set to null on insertion.

Google Calendar and Zoom OAuth tokens stored server-side are encrypted at rest using AES-256-GCM with keys derived from per-account random salts and the environment-level encryption key. Booking guest data (name and email) is encrypted using AES-256-GCM with per-row salts.

6.4 Licensing and Trial Data

Hyperdrive stores the following licensing-related data server-side:

  • Product records: Product identifier, name, maximum device limit, and grace period hours.
  • License records: HMAC-hashed license key (the plaintext key is not stored server-side), email address associated with the license, product identifier, license status (active, suspended, revoked, expired), maximum device override, expiration date, Stripe customer identifier, and Stripe subscription identifier.
  • Device activation records: License identifier, machine identifier (hardware UUID), machine name (e.g., "John's MacBook Pro"), platform (e.g., "macos"), application version, activation timestamp, and last validation timestamp.
  • Licensing audit event records: License identifier, event type, machine identifier, IP address of the API request, metadata JSON, and event timestamp. Event types include but are not limited to: activated, deactivated, validated, created, suspended, revoked, restored, and restore_failed.
  • Trial records: HMAC-hashed email address (the plaintext email is not stored), product identifier, machine identifier, IP address (truncated to /24 for IPv4 or /48 for IPv6), trial start timestamp, and trial duration in days.
  • License recovery codes: HMAC-hashed email address, product identifier, SHA-256 hashed verification code, attempt counter, creation timestamp, and expiration timestamp. Recovery codes expire after 10 minutes; used and expired codes are automatically deleted within 24 hours.
  • Rate limiting records: Rate limit key (IP-based, derived from the Cloudflare cf-connecting-ip header), sliding window start timestamp, and request count. IP addresses are used as rate-limit keys and stored transiently in a rate_limits table with short rolling windows.

6.5 AI Usage Metering

Hyperdrive logs the following information server-side for each AI feature request processed through the server proxy:

  • Account identifier (your email address, resolved server-side from your license, not taken from client headers).
  • Feature identifier (e.g., draft_reply, summarize_thread, summarize_slack, extract_calendar_event, rewrite_email, translate_search, smartVector, classify_senders, classify_pitch_body). The classify_senders feature is used for sender classification when senders have no cached or heuristic classification (see Section 10, Layer 5). The classify_pitch_body feature is used for cold pitch detection when heuristic scoring is ambiguous (see Section 10.2).
  • Model name used for the request.
  • Prompt token count, completion token count, and total token count.
  • Estimated cost in USD.
  • Request timestamp.

Important: AI usage metering logs the metadata about AI requests (which features were used, how many tokens were consumed) but does not log the content of AI requests or responses. The actual email content, prompts, and AI-generated text are not stored server-side.

AI usage metering data is currently stored without a defined expiration period. This data can be queried via an authenticated API endpoint.

6.6 Synced Snippets

Hyperdrive stores the following data server-side for each synced text snippet:

  • Snippet identifier (UUID).
  • Account identifier.
  • Snippet name (stored in plaintext).
  • Snippet body content (encrypted at rest using AES-256-GCM).
  • Text expansion shortcut, if configured (stored in plaintext).
  • Deleted flag (soft-delete indicator).
  • Creation timestamp and last-updated timestamp.

Retention: Synced snippets are retained until deleted by the user or until account disconnection. Soft-deleted snippets (deleted flag = 1) are purged after 30 days. All snippet data for an account is permanently deleted when the account is disconnected via the self-service account deletion endpoint.

6.7 Synced Signatures

Hyperdrive stores the following data server-side for each synced email signature:

  • Account identifier (one signature record per account).
  • Source identifier (e.g., gmail).
  • HTML body (encrypted at rest using AES-256-GCM).
  • Plain-text body (encrypted at rest using AES-256-GCM).
  • Last-updated timestamp.

Retention: Synced signatures are retained until replaced by the user or until account disconnection. All signature data for an account is permanently deleted when the account is disconnected via the self-service account deletion endpoint.


7. How Hyperdrive Uses Information

Hyperdrive uses the information it collects for the following purposes:

Core email client functionality:

  • Syncing email from Gmail via the Gmail API, including incremental sync via the History API.
  • Rendering, displaying, searching, and organizing emails locally.
  • Composing, drafting, sending, forwarding, and replying to emails.
  • Managing email labels, archiving, starring, trashing, marking as read/unread, and other email state changes.
  • Providing full-text search across locally cached email content.
  • Delivering macOS notifications for new emails when enabled.

Inbox organization and productivity:

  • Applying inbox vectors, automated classifications, and local heuristics to organize emails into categories.
  • Syncing sender classifications, thread vector overrides, and vector configurations across your devices via server-side storage.
  • If you opt in: contributing privacy-protected classification votes to the shared classification intelligence system using local differential privacy (see Section 10).
  • Providing snooze (remind me), reminders, and scheduled sending functionality, including server-side execution of due items when your device is offline.
  • Providing blocked sender management and unsubscribe detection.
  • Muting conversations.

Slack integration:

  • Syncing Slack messages and channel data when you connect a Slack workspace.
  • Displaying Slack messages alongside email.
  • Enabling Slack message sending and reactions where authorized by granted scopes.
  • Delivering macOS notifications for Slack messages when enabled.

Scheduling:

  • Providing scheduling link creation, management, and public booking pages.
  • Creating Google Calendar events when guests book meetings through your scheduling links.
  • Querying Google Calendar for free/busy information to present available time slots.

Calendar integration:

  • Syncing calendar events from Google Calendar.
  • Displaying calendar events in the calendar sidebar.
  • Creating, updating, and managing calendar events, including AI-assisted event extraction from email content.
  • Providing RSVP functionality for calendar invitations.
  • Sharing availability with email recipients.

Contacts integration:

  • Syncing your Google Contacts via the People API to provide recipient autocomplete in the email composer.
  • Reading your macOS system address book (Contacts framework) to supplement autocomplete results. macOS Contacts data is stored locally and never transmitted to any server.
  • Merging and ranking contacts from Google Contacts, macOS Contacts, and email interaction history to present the most relevant autocomplete suggestions.
  • Allowing you to create new contacts in your Google account via the “Add to Contacts” command.
  • Displaying your Google Contacts in an in-app panel via the “View Contacts” command, which loads contacts.google.com in an embedded web view with navigation restricted to Google domains.

AI-powered features:

  • Generating draft replies to emails.
  • Summarizing email threads and Slack channels.
  • Extracting calendar events from email content.
  • Rewriting and adjusting the tone of email drafts.
  • Translating natural language search queries into structured search parameters.
  • Classifying sender email addresses to support inbox vector organization.
  • Detecting cold pitch and sales outreach emails.
  • Generating smart vector filter predicates from natural language descriptions.

Contact enrichment:

  • Enriching contact information with publicly available profile data from Gravatar, including names, titles, companies, locations, photos, and social profile links.

Email tracking:

  • Detecting when recipients open tracked emails and click tracked links.
  • Providing open and click reporting to the sender.

Licensing and subscription management:

  • Enforcing license terms, device activation limits, and subscription status.
  • Managing trial periods.
  • Processing subscription payments through Stripe.
  • Preventing fraud and unauthorized use.

Security, integrity, and operations:

  • Maintaining the security and integrity of the Service.
  • Preventing abuse, enforcing rate limits, and detecting anomalous activity.
  • Monitoring server infrastructure for operational issues.
  • Verifying the integrity of licensing responses using cryptographic signatures.

Communications:

  • Sending transactional emails, including license confirmations, recovery codes, device activation notifications, and support responses.
  • Sending product communications, including download links, product updates, and feature announcements, to users who have provided their email address through the website, waitlist, trial signup, or license purchase.
  • You may unsubscribe from product communications at any time via the unsubscribe link included in every such email.

8. Legal Bases for Processing

Hyperdrive processes personal information under one or more of the following legal bases, depending on the specific processing activity and applicable jurisdiction:

Performance of a contract (Article 6(1)(b) GDPR):

  • Processing necessary to provide the Hyperdrive service you have requested, including email synchronization, sending, calendar integration, and account management.
  • Processing necessary to manage your license, trial, or subscription.

Legitimate interests (Article 6(1)(f) GDPR):

  • Securing the Service and preventing abuse, fraud, and unauthorized access.
  • Enforcing rate limits and maintaining service stability.
  • Operating licensing and device activation systems.
  • Improving the performance and reliability of the Service.
  • Metering AI usage for cost management and service planning.
  • Embedding email tracking pixels in outgoing emails when tracking is enabled (enabled by default; can be disabled in settings) and recording open and click events on Hyperdrive servers.
  • Querying Gravatar for publicly available contact profile data when viewing a contact’s profile in the application.

When relying on legitimate interests, Hyperdrive has conducted a balancing assessment and concluded that these interests are not overridden by your fundamental rights and freedoms. You have the right to object to processing based on legitimate interests (see Section 20).

Consent (Article 6(1)(a) GDPR):

  • Where required, such as when you affirmatively enable integrations and features that transmit data to third-party services (e.g., connecting a Slack workspace, enabling AI features that send email content to OpenAI through the server proxy).
  • Sending product communications (download links, product updates, feature announcements) to users who have affirmatively provided their email address through the website waitlist or signup form. You may withdraw this consent at any time by clicking the unsubscribe link in any product communication email.

Where processing is based on consent, you have the right to withdraw your consent at any time. Withdrawal of consent does not affect the lawfulness of processing carried out before the withdrawal.

Compliance with legal obligations (Article 6(1)(c) GDPR):

  • Processing necessary to comply with applicable legal obligations, such as responding to lawful requests from law enforcement or regulatory authorities.
  • Maintaining records as required by tax, accounting, or other regulatory requirements.

Because Hyperdrive processes sensitive communications data, including the content of private emails and messages, you should use Hyperdrive only if you have the legal authority and right to access, process, and transmit the content you access through the Service.


9. Third-Party Services and Data Sharing

Hyperdrive interacts with multiple third-party services to provide its functionality. When you enable features that involve third-party services, data is transmitted to those services as necessary. Hyperdrive does not sell, rent, or trade your personal information to third parties for their own marketing purposes.

9.1 Google (Gmail API, Calendar API, and People API)

When you connect a Google account, Hyperdrive communicates with Google’s Gmail API, Calendar API, and People API over HTTPS using OAuth 2.0 authentication. The OAuth authorization flow uses a system-managed secure browser session.

OAuth scopes requested:

  • gmail.modify — Read and modify email, labels, and threads.
  • gmail.compose — Compose, draft, and send emails.
  • gmail.settings.basic — Read and modify basic Gmail settings.
  • userinfo.email — Access your email address.
  • calendar — Read/write access to Google Calendar events and free/busy queries.
  • contacts — Read/write access to Google Contacts for autocomplete and the “Add to Contacts” feature.

Data transmitted to Google may include:

  • Email content and metadata for reading, searching, drafting, and sending.
  • Label modifications and email state changes.
  • Calendar events created, updated, or deleted, and free/busy queries.
  • Search queries for remote search.
  • Contact read requests (names, email addresses, phone numbers, organizations, and group memberships) via the People API for autocomplete. New contact records (name and email) when you use “Add to Contacts.”

9.2 Slack API

If you connect a Slack workspace, Hyperdrive communicates with the Slack API over HTTPS. The Slack OAuth token exchange is handled server-side through Hyperdrive servers to keep credentials out of the client application.

OAuth scopes requested: Hyperdrive requests read, write, and history access to public channels, private channels, direct messages, and group direct messages; permission to send messages and post reactions on your behalf; and access to user profile information and email addresses.

Data transmitted to Slack may include:

  • Message read requests across all channel types, depending on granted scopes.
  • Messages you send or reactions you post through Hyperdrive.
  • User profile and email address lookups.

Slack user tokens do not expire and remain valid until you revoke them through Slack’s app management interface or disconnect the workspace in Hyperdrive.

9.3 OpenAI (via Server Proxy)

If you use cloud-based AI features, Hyperdrive sends data to Hyperdrive servers, which forward requests to OpenAI’s API. The OpenAI API key is stored server-side and is never present in the client application. On-device AI features do not use the server proxy and do not transmit data to OpenAI. The specific data transmitted depends on the feature and is detailed in Section 10.

Hyperdrive servers log metadata about AI requests (account identifier, feature name, model used, token counts, and estimated cost) but do not log or store the content of prompts, email content, or AI-generated responses. Email content passes through the server proxy transiently and is not persisted.

Custom AI tier: If you provide your own API key and optional custom endpoint URL, Hyperdrive sends AI requests directly to the endpoint you configure, bypassing the Hyperdrive proxy. In this configuration, Hyperdrive does not receive, process, or meter your AI requests. You are solely responsible for evaluating the privacy and security practices of any custom endpoint you configure.

9.4 Gravatar (Contact Enrichment)

For contact enrichment, Hyperdrive sends a cryptographic hash of the contact’s email address to the Gravatar API. The email address itself is not sent — only its hash. Gravatar returns publicly available profile information, which may include: name, company, title, location, photo URL, and social profile links.

Contact enrichment data is encrypted at rest and cached server-side for up to 30 days before automatic expiration.

Gravatar’s operator (Automattic) may be able to correlate the hash with known email addresses if those addresses are registered in the Gravatar system.

9.5 Unsplash (Inbox Zero Photos)

Hyperdrive displays photographs from Unsplash for the inbox-zero celebration feature. The Hyperdrive server fetches and caches a daily photo from the Unsplash API. Your device downloads the image directly from Unsplash’s CDN, which exposes your IP address to Unsplash’s infrastructure. No personally identifiable user data is included in requests to Unsplash.

9.6 Stripe (Payments)

If you purchase a Hyperdrive subscription, payment processing is handled entirely by Stripe. Hyperdrive does not receive, process, or store your payment card details, bank account information, or other payment instrument data.

Hyperdrive receives and stores Stripe customer, subscription, and payment intent identifiers for subscription management. Stripe notifies Hyperdrive of subscription lifecycle events (creation, cancellation, updates) via authenticated webhooks.

After successful payment, a license confirmation email is sent to you via Resend, a third-party email delivery service. The confirmation email does not contain your license key. Resend receives your email address and the email content for delivery purposes; Resend’s privacy policy governs their handling of this data.

9.7 Cloudflare (Infrastructure)

Hyperdrive servers run on Cloudflare infrastructure. As part of providing its services, Cloudflare may process IP addresses, request metadata, and timing information. Cloudflare processes data at global edge locations; Cloudflare’s privacy policy governs its handling of this data.

9.8 HuggingFace (Model Download)

On Apple Silicon Macs, Hyperdrive performs a one-time download of an on-device language model from HuggingFace. Only standard HTTP request headers (including your IP address) are transmitted; no email content, account information, or personal data is sent. After the initial download, the model operates entirely offline. HuggingFace’s privacy policy governs their handling of download request metadata.

9.9 Sparkle (Software Updates)

Hyperdrive uses the open-source Sparkle framework for automatic software updates. Sparkle periodically checks for updates from Hyperdrive’s server. Data transmitted is limited to standard HTTP request headers (including app version, macOS version, and your IP address). No system profiling or hardware inventory data is sent. Update binaries are cryptographically verified before installation.

9.10 Services Hyperdrive Does NOT Use

Hyperdrive does not integrate with or transmit data to:

  • Advertising networks or ad SDKs.
  • Analytics platforms.
  • Crash reporting services.
  • iCloud or Apple’s cloud synchronization services.
  • Data brokers or data resellers.

10. Artificial Intelligence and Machine Learning Features

Hyperdrive offers AI-powered features that use a combination of on-device machine learning models and cloud-based language models (OpenAI, accessed through the Hyperdrive server proxy, or your custom endpoint if configured).

On-device AI features process data entirely on your Mac using bundled or locally downloaded ML models. No email content leaves your device for these features. On-device AI features include email classification (sender categorization, cold pitch detection, and inbox vector assignment).

Cloud AI features transmit portions of your email or Slack content to OpenAI for processing through the Hyperdrive server proxy. Cloud AI features include draft reply generation, email rewriting, thread and Slack summarization, calendar event extraction, search query translation, smart vector predicate generation, cloud sender classification, and ambiguous cold pitch body classification.

Cloud sender classification: When Hyperdrive encounters unknown senders not yet in the local classification cache, it may send sender email addresses and up to 3 sample subject lines per sender to OpenAI via the server proxy for batch classification. On-device models and heuristics handle the majority of classifications without transmitting any data. Email bodies are not sent for batch sender classification.

Cold pitch body classification: For emails where on-device heuristic scoring produces an ambiguous cold pitch result (score 2–3 out of 10+), Hyperdrive may transmit up to 500 characters of de-identified email body text to OpenAI via the server proxy for a definitive pitch determination. Before transmission, personally identifiable information is removed: email addresses are replaced with [email], phone numbers with [phone], URLs with [url], and email signatures are stripped. The sender’s domain (not full email address) and the email subject line are also sent. This feature is invoked only for emails from senders you have not previously corresponded with, who have sent fewer than 5 emails, and whose message format and content match cold outreach patterns. The result (cold_pitch or other) is cached locally and synced across devices. Feature identifier: classify_pitch_body. Model: GPT-4.1-mini.

Consent: Draft reply generation, email rewriting, thread and Slack summarization, calendar event extraction, and search query translation are gated by the AI consent panel (Settings → Privacy); disabling AI in settings prevents these features from functioning. Smart vector predicate generation requires explicit user action to invoke but is not gated by the AI consent panel. Cloud sender classification and cold pitch body classification run automatically when AI features are enabled. On-device AI features run automatically without a separate consent gate, as no data leaves your device.

10.1 Cloud AI Features — Data Sent per Feature

When you use a cloud AI feature, Hyperdrive sends limited portions of your content to the AI provider through the server proxy. The following summarizes what data is sent for each feature:

  • Draft reply generation: The last 3 messages from the email thread, each including the sender’s display name and the message body truncated to 2,000 characters.
  • Thread summarization: The last 10 messages from the email thread, each including the sender’s display name (or email address if no display name is available) and message body text truncated to 2,000 characters. Generated summaries are cached locally so the same thread is not re-summarized on subsequent views.
  • Slack channel summarization: The last 50 messages from the selected channel, each including the sender’s display name (or email address if unavailable) and message text truncated to 1,000 characters, along with the channel name.
  • Calendar event extraction: The last 2 messages from the email thread, each including the sender’s display name (or email address if unavailable), subject line, date, and the last 2,000 characters of the message body. When extracting from a draft you are composing, the subject line and email body truncated to 3,000 characters are sent.
  • Email rewriting and tone adjustment: The full email body text you are composing, your selected tone preference, and any optional custom instructions you provide.
  • Natural language search translation: Your search query text only. No email content is sent.

10.2 Email Classification Pipeline

Hyperdrive classifies emails into inbox vectors using a multi-layer pipeline that prioritizes on-device processing. Most emails are classified without any data leaving your device.

On-device processing (no data transmitted): Your manual overrides always take precedence. Deterministic rules, heuristic scoring, and bundled machine learning models classify the majority of emails entirely on-device. On Apple Silicon Macs, a locally downloaded language model provides additional classification for ambiguous cases.

Cloud classification (data transmitted to the AI provider via server proxy): For senders not yet in the local classification cache, Hyperdrive may send sender email addresses and a small number of sample subject lines per sender in batches. No email body text is sent for batch sender classification. For cold pitch detection, when on-device heuristic scoring is ambiguous, up to 500 characters of de-identified email body text (with email addresses, phone numbers, URLs, and signatures removed) may be sent for a definitive classification. See Section 10 for full details on cold pitch body classification data handling.

Local caching: All classification results are cached locally indefinitely.

10.3 Smart Vector Predicate Generation

When you create a smart vector, Hyperdrive sends your natural language filter description and a sample of thread metadata (sender addresses, subject lines, dates, and short snippets) to the AI provider to generate a filter predicate. During refinement, only the current filter definition and your instructions are sent — no email content. Generated predicates are evaluated entirely locally against thread metadata thereafter.

10.4 Shared Classification Intelligence (Opt-In)

Hyperdrive offers an optional, opt-in feature that improves classification accuracy by aggregating anonymized sender category votes across participating users.

What is shared: When you opt in, your device periodically contributes sender category votes. Only a cryptographically hashed sender identifier (using a periodically rotating salt) and a category vote with differential privacy noise applied on-device are transmitted. Raw email addresses are never transmitted. Email content, bodies, subject lines, recipient information, and thread data are never shared.

Privacy protections: Differential privacy provides a mathematical guarantee that individual contributions cannot be determined from aggregate data. Results are only returned once a minimum vote threshold is reached per sender. This feature is disabled by default and requires explicit opt-in via Settings → Privacy → Shared Classification Intelligence. You may opt out at any time. Previously contributed votes cannot be individually withdrawn due to differential privacy aggregation.


11. Email Tracking and Recipient Data

Hyperdrive includes a feature that injects a tracking pixel (a 1-pixel transparent GIF image) and wraps hyperlinks in outbound emails to enable detection of email opens and link clicks.

11.1 How Tracking Works

  1. Pixel injection: When you send an email with tracking enabled, Hyperdrive appends an invisible 1x1 pixel image to the HTML body of the email.

  2. Link wrapping: All hyperlinks in the outbound email are rewritten to route through the Hyperdrive server, which records the click and redirects the recipient to the original destination.

  3. Pixel and link registration: When you send a tracked email, Hyperdrive registers the pixel and tracked links with the server. Registered data includes your account identifier, recipient email address, email subject, thread identifier, and destination URLs.

  4. Open detection: When a recipient’s email client loads the tracking pixel, the server logs an open event with a truncated IP address, user-agent string, and timestamp.

  5. Click detection: When a recipient clicks a tracked link, the server logs the click event (truncated IP, user-agent, timestamp, and destination URL) and redirects to the original URL.

  6. Reporting: Hyperdrive periodically polls the server for open events and displays a recent-opens feed in the contact pane.

11.2 Data Collected About Email Recipients

Email tracking collects the following information about recipients of your tracked emails. Recipients are third parties who have not agreed to this Privacy Policy, and this data is collected without their explicit knowledge or consent:

  • Recipient email address (provided by you when sending).
  • Truncated IP address of the recipient's device or email client at the time of each open or click event (last octet zeroed for IPv4, /48 prefix for IPv6).
  • User-agent string of the recipient's email client or browser, which may reveal device type, operating system, and software version.
  • Timestamps of each open and click event.
  • Which specific links were clicked and their destination URLs.

11.3 Recipient Notice and Consent

Hyperdrive does not automatically notify recipients that email tracking is active. The tracking pixel and link wrapping are designed to be invisible to recipients.

You can disable email tracking at any time via the “Enable read receipt tracking” toggle in Settings → Privacy. When tracking is disabled, no tracking pixel is injected and no links are wrapped in outgoing emails. Tracking is enabled by default.

You are solely responsible for complying with all laws, regulations, and policies that govern email tracking, monitoring, consent, and notice in your jurisdiction and the jurisdiction of your recipients. This may include, without limitation:

  • Laws requiring recipient consent before tracking (such as GDPR in the European Union, which may require consent for tracking technologies).
  • Employer policies governing email monitoring of employees.
  • Confidentiality obligations that may prohibit tracking of communications with attorneys, medical professionals, or other privileged parties.
  • Industry-specific regulations governing electronic communications.

Hyperdrive is a tool. Hyperdrive does not provide legal advice, does not guarantee compliance with any particular law or regulation, and is not responsible for your use of tracking features in violation of applicable law.


12. Automated Decision-Making and Profiling

Hyperdrive employs automated classification and decision-making systems that affect which emails you see, how they are organized, and how they are prioritized. While these systems are designed to improve your productivity, you should be aware of the following automated processes:

12.1 Inbox Vector Classification

Hyperdrive automatically classifies incoming emails into inbox vectors using a combination of on-device models, heuristics, and — for unknown senders not resolved on-device — cloud-based classification. Your manual overrides always take precedence. The majority of classifications are resolved entirely on-device without any network communication. See Section 10.2 for details on what data is transmitted for cloud classification.

These classifications determine which inbox vector a given email appears in and may affect the prominence and visibility of emails in your inbox.

12.2 Cold Pitch Detection

Hyperdrive automatically detects unsolicited sales and outreach emails ("cold pitches") using on-device heuristic analysis. When on-device scoring produces an ambiguous result, up to 500 characters of de-identified email body text may be transmitted to an AI provider for a definitive classification. Emails classified as cold pitches may be de-prioritized or categorized differently in your inbox. See Section 10 and Section 10.2 for details on what data is transmitted.

12.3 Your Rights Regarding Automated Decision-Making

Under GDPR Article 22, you have the right not to be subject to decisions based solely on automated processing that produce legal effects or similarly significantly affect you. Inbox classification does not produce legal effects, but it may affect which communications you see promptly. You can override any automated classification by manually moving emails between vectors or adjusting vector rules.


13. Local-Only Processing

Certain Hyperdrive features are performed entirely locally on your Mac without sending any content or data to Hyperdrive servers or any third-party service. These include:

  • Cold pitch heuristic detection using locally computed indicators (ambiguous results may be escalated to cloud AI; see Section 10).
  • On-device machine learning classification using bundled and locally downloaded models.
  • Multi-layer vector classification and rule evaluation across all on-device pipeline layers.
  • Smart vector predicate evaluation against local email metadata.
  • Remote image blocking and whitelist enforcement.
  • Full-text search indexing and query execution.
  • Email HTML sanitization before rendering.
  • Draft auto-saving.
  • Optimistic UI updates for email actions.

14. Hardware Identifiers, Device Fingerprinting, and Licensing

14.1 Hardware UUID Collection

Hyperdrive licensing uses a persistent hardware identifier derived from your Mac’s platform UUID. This UUID is unique to each physical Mac and does not change across operating system reinstalls, user account changes, or application reinstalls. If the platform UUID is unavailable, Hyperdrive falls back to a randomly generated identifier that is not persisted across launches.

14.2 Data Collected for Licensing

Licensing operations transmit data to the licensing server at app.hyperdriveone.com. The exact fields vary by operation:

License activation, validation, and deactivation:

  • License key — sent on activation, validation, and deactivation.
  • Machine identifier (hardware UUID as described above) — sent on activation, validation, and deactivation.
  • Product and platform identifiers — sent on activation and validation.
  • Application version — sent on activation and validation.
  • Machine name (your Mac’s user-assigned name, which may contain your name) — sent on activation only.

Trial start and validation: Your email address, product identifier, and machine identifier are transmitted. Your email is stored only as an irreversible hash on the server.

Checkout (purchase): Your email address, product identifier, selected plan, machine identifier, and machine name are transmitted.

License recovery: Your email address and a time-limited verification code are transmitted. Recovery codes expire after 10 minutes and are automatically deleted within 24 hours.

The combination of hardware UUID, machine name, and email address constitutes a device-level fingerprint. A hashed fingerprint is stored server-side for device identity verification.

14.3 Licensing Enforcement

  • License activation is validated approximately every 24 hours via a server API call.
  • A configurable grace period allows offline use when the validation server is unreachable.
  • Server responses are cryptographically signed to prevent tampering.
  • Rate limits are enforced on licensing endpoints to prevent abuse.
  • Device limits are enforced per license.

14.4 Licensing Event Audit Log

The licensing server maintains an audit log of licensing events, recording the license identifier, event type, machine identifier, truncated IP address, and timestamp. This audit log is retained indefinitely.


15. Notifications

If you grant notification permission, Hyperdrive may display macOS system notifications that include:

  • Email notifications: Email subject line and a short message snippet. Scheduled-send confirmations display the recipient address.
  • Slack notifications: Thread subject or “Slack message” and a message snippet.
  • Reminder notifications: Thread subject for reminded conversations.

Notification content is visible on your screen (including the lock screen, depending on your macOS notification settings) even when the Hyperdrive application is not in the foreground. You can control notification display behavior at two levels:

  1. Within Hyperdrive: Settings provide individual toggles for Gmail notifications, Slack notifications, reminder notifications, and notification sound.
  2. Within macOS System Settings: You can manage notification permissions, display style, and lock screen visibility for Hyperdrive at the operating system level.

16. Data Retention

Hyperdrive's data retention practices depend on where data is stored and the type of data.

16.1 Local Storage Retention

  • Email and Slack cache: Data stored in the local SQLite database persists until you delete it, remove accounts, or uninstall the application. A daily pruning job automatically removes cached threads that are no longer in your inbox and are older than 30 days. Inbox threads and their associated messages are not automatically pruned.
  • Keychain credentials: OAuth tokens and other credentials stored in the macOS Keychain persist until you disconnect accounts, explicitly clear credentials, or uninstall the application. Keychain items with the kSecAttrAccessibleWhenUnlockedThisDeviceOnly attribute are also removed if you erase the device.
  • UserDefaults preferences: Preferences persist until you reset them within the application or delete the preference file at ~/Library/Preferences/com.hyperdrive.app.plist.
  • Search history: Recent search queries (up to 10) persist until manually cleared or until the preference store is reset.
  • Vector configuration: JSON configuration files persist until you modify or delete them.

16.2 Server-Side Retention with Defined Periods

Data Retention Period Mechanism
Contact enrichment cache 30 days expires_at column; records refreshed on re-query after expiry
Sender classification cache (local) 30 days Local expiry; re-classified via on-device classification pipeline after expiry
Unsplash photo cache (local) 24 hours Refreshed via midnight timer and system wake events; server rotates photo daily
Rate limiting counters 1–1,440 minute sliding windows (varies by endpoint) Automatically recycled; stale records cleaned after 1 hour
Email tracking pixel registrations 90 days Automated deletion via automated retention cleanup; pixel registration records, open events, and click events older than 90 days are permanently deleted
Email open events 90 days Automated deletion via automated retention cleanup
Link click events 90 days Automated deletion via automated retention cleanup
Stale device registrations 90 days Automated deletion via automated retention cleanup; devices not seen in 90 days are removed
Admin access logs 90 days Automated deletion via automated retention cleanup
Completed scheduled messages 7 days (sent) / 30 days (failed or cancelled) Automated deletion via automated retention cleanup; associated R2 attachments are also deleted
Completed snoozes and reminders 30 days Automated deletion via automated retention cleanup after snooze has fired or been cancelled, or after reminder has fired, been skipped, or been cancelled
Synced thread overrides 365 days Automated deletion via automated retention cleanup
License recovery codes 10 minutes (expiry) / 24 hours (used codes) Automated deletion via automated retention cleanup; expired codes and codes used more than 24 hours ago are permanently deleted
Synced snippets Until deleted by user or account disconnection Deleted on user action or account disconnection via self-service account deletion endpoint
Soft-deleted snippets (deleted flag = 1) 30 days Purged after 30 days from soft-deletion
Synced signatures Until replaced by user or account disconnection Deleted on user action or account disconnection via self-service account deletion endpoint
Stripe webhook idempotency records 7 days Automated deletion via automated retention cleanup

16.3 Server-Side Data Retained Indefinitely

The following categories of server-side data do not have a defined expiration or automatic deletion schedule and are retained indefinitely:

Category What is retained Justification
Licensing and billing records License records (email address, HMAC-hashed license key, Stripe identifiers, license status, timestamps), trial records (HMAC-hashed email, machine identifier, truncated IP address), and device activation records. License and trial records are not deleted by the self-service account data deletion endpoint. Fraud prevention, subscription management, and legal compliance
Audit and security logs Licensing event logs (event type, machine identifier, truncated IP address, timestamps) and application download logs (truncated IP address, user agent, timestamp). Security monitoring and dispute resolution
AI usage metering Account-level feature usage patterns (feature name, model, token counts, estimated cost, timestamp). Content of AI requests is not retained. Cost management and service planning
Shared Classification Intelligence votes (opt-in only) Device identifier, cryptographically hashed sender token, differentially private noised category, and salt epoch. Votes are not deleted by the self-service account data deletion endpoint and cannot be individually withdrawn due to differential privacy aggregation (see Section 10.4). Aggregate classification accuracy across participating users

Hyperdrive may implement retention limits for these categories in the future. Until such measures are implemented, data is retained as long as reasonably necessary to provide the Service, comply with legal obligations, resolve disputes, enforce agreements, and maintain security and operational integrity.


17. Data Deletion, Export, and Account Controls

17.1 Current Capabilities

At present, the following user-initiated data management capabilities are available:

  • Local data removal: You can remove local data by uninstalling Hyperdrive and deleting the Hyperdrive data directories at ~/Library/Application Support/Hyperdrive/ and ~/Library/Caches/Hyperdrive/, and the preferences file at ~/Library/Preferences/com.hyperdrive.app.plist. If you downloaded the on-device MLX model, also delete the HuggingFace hub cache (see On-device model management below). Keychain items are not automatically removed when you delete the application; disconnect accounts within Hyperdrive before uninstalling, or manually remove items via Keychain Access.
  • Account disconnection: You can disconnect email and Slack accounts within Hyperdrive, which removes the associated OAuth tokens from the local Keychain and clears local cached data for that account.
  • License deactivation: You can deactivate your license on a specific device via Settings → Subscription → Deactivate This Device, which calls the licensing server (POST /api/licenses/deactivate) to remove the device activation record for that machine.
  • Image whitelist management: You can add, remove, or clear sender and domain entries from the remote image whitelist.
  • Blocked sender management: You can block and unblock senders.
  • Search history: You can clear recent search queries.
  • Server-side data deletion: You can delete all server-side data associated with an email account (including snoozed threads, reminders, scheduled messages and attachments, tracking data, AI usage logs, Gmail credentials, device registrations, restore codes, synced sender classifications, synced thread overrides, synced vector configurations, scheduling links, bookings, and scheduling user profiles) via Settings → Accounts → Delete Server Data or by removing the account. This operation is also available via the server API (DELETE /api/account). The license record itself is preserved, but license activations and audit events are deleted.
  • Cloud AI feature toggle: You can enable or disable cloud-based AI features (draft replies, summaries, rewrites, calendar event extraction, search translation, sender classification, and smart vector generation) via Settings → Privacy → AI Features. When disabled, no email content is sent to OpenAI. On-device classification continues to operate locally regardless of this toggle. A one-time consent dialog is shown on first launch.
  • On-device model management: You can delete the downloaded on-device language model via Settings. Model files are stored in a system-managed cache under ~/Library/Caches/. If deleted, classification falls back to other classification methods and, if needed, cloud classification.
  • Shared Classification Intelligence opt-out: If opted in, you can opt out at any time via the “Contribute anonymous sender categories” toggle in Settings → Privacy. Opting out stops contribution of new votes. Previously contributed votes cannot be individually withdrawn due to differential privacy.
  • Email tracking toggle: You can enable or disable read receipt tracking via the “Enable read receipt tracking” toggle in Settings → Privacy. When disabled, no tracking pixel is injected and no links are wrapped in outgoing emails.
  • Data export: You can export all server-side personal data for any connected account via Settings → Accounts → Export Data. The export is a JSON file containing your snoozes, reminders, scheduled messages, tracking data, AI usage logs, device registrations, synced classifications, synced thread overrides, synced vector configuration, synced snippets, synced signatures, scheduling links, and booking data. Encrypted fields are decrypted at export time. OAuth credentials are excluded for security. This operation is also available via the server API (GET /api/account/export).

17.2 Current Limitations

At present, the following capabilities are not available:

  • No toggle to opt out of AI usage metering. AI usage metering is logged automatically for all AI feature requests processed through the server proxy.

17.3 Requesting Deletion or Export

You can export your server-side data at any time using the Export Data button in Settings → Accounts. For server-side data deletion, use the Delete Server Data button in the same location, or contact Hyperdrive at the contact details provided in Section 28.

Hyperdrive will evaluate requests consistent with applicable law, including GDPR, CCPA/CPRA, and other data protection regulations. You may be asked to verify your identity before information is provided, modified, or deleted. Hyperdrive will respond to verified requests within the timeframes required by applicable law (generally 30 days under GDPR, 45 days under CCPA).


18. Security Measures

Hyperdrive implements a range of technical and organizational measures designed to protect your information. However, no system can be guaranteed to be completely secure, and Hyperdrive cannot guarantee the absolute security of your data.

18.1 Transport Security

  • All network communications between the Hyperdrive client and Hyperdrive servers, as well as between Hyperdrive servers and third-party APIs, use HTTPS (TLS encryption in transit). No unencrypted HTTP endpoints are used.

18.2 Credential Storage

  • OAuth tokens and sensitive credentials on the client are stored in the macOS system Keychain using kSecAttrAccessibleWhenUnlockedThisDeviceOnly, which restricts access to when the device is unlocked and prevents iCloud Keychain synchronization. Keychain items are encrypted by macOS using the login keychain’s password-derived encryption.
  • OAuth tokens stored server-side (Gmail tokens for deferred sync features, and Google Calendar and Zoom tokens for scheduling) are encrypted at rest using AES-256-GCM with keys derived from per-account salts and environment-specific secrets.
  • Scheduling API tokens and deferred-action account tokens are hashed (SHA-256) before storage; plaintext tokens are never persisted.

18.3 Cryptographic Verification

  • Licensing server responses are digitally signed using Ed25519 (a modern elliptic curve signature algorithm). The Hyperdrive client verifies these signatures using a hardcoded public key, preventing response tampering or replay attacks.

18.4 Authentication and Access Control

  • Server-side administrative endpoints for licensing are protected by HMAC-based authentication with timestamp-based replay protection (2-minute validity window).
  • Stripe webhook payloads are verified using HMAC-SHA256 signature verification with a server-side webhook secret. Stripe webhook event IDs are stored server-side for idempotency and replay prevention.
  • Slack OAuth state parameters are validated to prevent CSRF attacks.
  • Gmail OAuth uses ASWebAuthenticationSession, which provides system-managed CSRF protection.

18.5 Email Content Safety

  • Hyperdrive sanitizes all HTML email content before rendering in WKWebView, stripping dangerous elements including <script>, <iframe>, <object>, <embed>, <form>, <input>, <base>, and all on* event handler attributes.
  • Email-originated JavaScript is blocked in rendering web views via Content Security Policy (default-src 'none' with no script-src directive) and HTML sanitization (stripping <script> tags).
  • Content Security Policy headers restrict resource loading in email views.
  • Navigation is blocked in email web views to prevent link hijacking.

18.6 Rate Limiting

  • Per-IP rate limits are enforced on all public licensing endpoints. Administrative endpoints are protected by HMAC authentication.
  • Slack API retry logic implements Retry-After header support and exponential backoff.

18.7 Your Security Responsibilities

You are responsible for:

  • Protecting physical access to your Mac and securing your macOS user account with a strong password.
  • Keeping your macOS operating system and Hyperdrive application updated.
  • Using appropriate endpoint security practices (such as FileVault disk encryption).
  • Protecting your license key and not sharing it with unauthorized users.
  • Revoking OAuth access from connected services if you suspect unauthorized access (via Google Account settings, Slack app management, etc.).

19. International Data Transfers

Hyperdrive involves international data processing. If you are located outside the United States, your data may be transferred to, stored in, and processed in the United States or other countries. Specifically:

  • Cloudflare Workers may process your requests at any of Cloudflare's global edge locations, depending on network routing and availability. The specific location may vary per request.
  • Cloudflare D1 stores data at edge locations selected by Cloudflare.
  • Cloudflare R2 stores attachment data at locations selected by Cloudflare.
  • OpenAI processes AI requests in regions determined by OpenAI's infrastructure, which may include the United States and other locations.
  • Gravatar (Automattic) processes contact enrichment data, primarily in the United States.
  • Stripe processes payment data in the United States and other locations, in compliance with PCI DSS.
  • Unsplash may process data globally.
  • HuggingFace serves model weight downloads from infrastructure that may be located in the United States, Europe, or other regions.
  • Google processes Gmail and Calendar data in regions determined by Google's infrastructure.
  • Resend delivers transactional and product communication emails (license confirmations, recovery codes, device activation notifications, waitlist confirmations, support form submissions, download links, product updates, and feature announcements) from infrastructure that may be located in the United States.
  • Slack processes workspace authentication and messaging data in regions determined by Slack's infrastructure, if you connect a Slack workspace.
  • Zoom processes OAuth authentication and meeting creation data in regions determined by Zoom's infrastructure, if you connect a Zoom account for scheduling link meetings.

For users in the European Economic Area (EEA), United Kingdom, or Switzerland: Transfers of personal data to countries outside the EEA that have not received an adequacy decision from the European Commission are conducted on the basis of appropriate safeguards, including Standard Contractual Clauses (SCCs) where applicable, or reliance on the recipient's participation in recognized frameworks. Hyperdrive does not currently maintain separate Data Processing Agreements (DPAs) with all third-party service providers and may not be able to guarantee that all sub-processors meet EEA adequacy requirements.

If you are located in a jurisdiction that restricts cross-border data transfers, you should evaluate whether use of Hyperdrive is appropriate for your regulatory requirements before using the Service.


20. Your Rights Under the General Data Protection Regulation (GDPR)

If you are located in the European Economic Area (EEA), the United Kingdom, or Switzerland, you have the following rights under the GDPR with respect to your personal data:

Right Description How to Exercise
Right of access (Art. 15) You have the right to obtain confirmation of whether we process your personal data and to access a copy of that data, along with information about the purposes, categories, recipients, retention periods, and safeguards. Contact us at the address in Section 28.
Right to rectification (Art. 16) You have the right to request correction of inaccurate personal data and completion of incomplete personal data. Contact us at the address in Section 28.
Right to erasure (Art. 17) You have the right to request deletion of your personal data where it is no longer necessary, where you withdraw consent, where you object to processing, where data was unlawfully processed, or where deletion is required by law. Contact us at the address in Section 28. See also Section 17.
Right to restriction (Art. 18) You have the right to request restriction of processing where accuracy is contested, processing is unlawful, we no longer need the data but you require it for legal claims, or you have objected to processing pending verification. Contact us at the address in Section 28.
Right to data portability (Art. 20) You have the right to receive your personal data in a structured, commonly used, and machine-readable format, and to transmit that data to another controller. Use the Export Data button in Settings → Accounts, or contact us at the address in Section 28.
Right to object (Art. 21) You have the right to object to processing based on legitimate interests or for direct marketing purposes. Contact us at the address in Section 28. You can also disable specific features (such as AI features, Slack integration, scheduling) by disconnecting services or avoiding use of those features.
Right not to be subject to automated decision-making (Art. 22) You have the right not to be subject to decisions based solely on automated processing that produce legal effects or similarly significantly affect you. See Section 12.3. You can override any automated inbox classification manually.
Right to withdraw consent Where processing is based on consent, you have the right to withdraw consent at any time without affecting the lawfulness of processing before withdrawal. Disconnect services, disable features, or contact us at the address in Section 28.
Right to lodge a complaint You have the right to lodge a complaint with a supervisory authority in the EU Member State of your habitual residence, place of work, or place of the alleged infringement. Contact your local data protection authority. A list of EU data protection authorities is available at the European Data Protection Board website.

Response time: Hyperdrive will respond to verified requests within 30 days, which may be extended by an additional 60 days for complex or numerous requests, in which case we will inform you of the extension and the reasons for it.

Identity verification: To protect your privacy, we may require you to verify your identity before acting on a request. This may include confirming ownership of the email address associated with your account or license.

Current implementation status: Hyperdrive provides self-service tools for data deletion (Section 17.1) and data export (Section 17.1). For other GDPR rights not covered by in-app controls, requests are handled manually via the contact details in Section 28.


21. Your Rights Under the California Consumer Privacy Act (CCPA/CPRA)

If you are a California resident, you have the following rights under the California Consumer Privacy Act, as amended by the California Privacy Rights Act (CCPA/CPRA):

21.1 Right to Know

You have the right to request that we disclose the categories and specific pieces of personal information we have collected about you, the categories of sources from which it was collected, the business or commercial purpose for collection, and the categories of third parties with whom we share it.

21.2 Right to Delete

You have the right to request deletion of your personal information, subject to certain exceptions (such as where retention is necessary to complete a transaction, detect security incidents, comply with legal obligations, or exercise free speech).

21.3 Right to Correct

You have the right to request correction of inaccurate personal information.

21.4 Right to Opt Out of Sale or Sharing

Hyperdrive does not sell your personal information and does not share your personal information for cross-context behavioral advertising purposes as those terms are defined under the CCPA/CPRA. Accordingly, there is no need to opt out of sale or sharing.

21.5 Right to Limit Use of Sensitive Personal Information

To the extent Hyperdrive processes sensitive personal information (such as email content), it does so only as necessary to provide the Service you have requested. Hyperdrive does not use sensitive personal information for purposes beyond what is necessary for the Service.

21.6 Right to Non-Discrimination

Hyperdrive will not discriminate against you for exercising any of your CCPA/CPRA rights.

21.7 Categories of Personal Information Collected

For purposes of CCPA disclosure, the following categories of personal information may be collected:

CCPA Category Examples Collected
Identifiers Email address, device identifier, IP address, license key Yes
Customer records Name, email address, subscription information Yes
Commercial information Subscription status, purchase history via Stripe Yes
Internet or electronic network activity Email tracking opens/clicks, AI feature usage, user-agent strings Yes
Geolocation data Approximate location via IP address (not precise GPS) Yes
Professional or employment-related information Job title, company (via Gravatar enrichment) Yes
Inferences drawn from personal information Inbox vector classifications, sender relationship data, AI classifications Yes
Sensitive personal information Email content, Slack message content, OAuth credentials Yes

21.8 How to Exercise Your Rights

To exercise any CCPA/CPRA rights, contact Hyperdrive at the address in Section 28. Hyperdrive will respond to verified requests within 45 days, which may be extended by an additional 45 days for complex requests.


22. Do Not Track Signals

Some web browsers transmit "Do Not Track" (DNT) signals to websites. Hyperdrive is a native macOS application and does not operate as a website visited in a browser, so DNT signals from browsers are not applicable to the Hyperdrive application.

For web-based scheduling pages served at app.hyperdriveone.com, Hyperdrive does not currently respond to or alter its practices upon receiving DNT signals, as there is no industry-standard technology for honoring DNT in this context.


23. Children's Privacy

Hyperdrive is not directed to children under the age of 16 (or such other age as may be specified by applicable law) and is intended for use by individuals who can lawfully form a binding contract. Hyperdrive does not knowingly collect, solicit, or process personal information from children under the age of 16.

If you are a parent or guardian and believe that a child under the age of 16 has provided personal information to Hyperdrive through Hyperdrive, please contact us at the address in Section 28, and we will take steps to delete such information from our systems.

In the United States, the Children's Online Privacy Protection Act (COPPA) applies to children under the age of 13. Hyperdrive does not knowingly collect personal information from children under 13 in the United States.


24. Information We Do Not Collect

For transparency, Hyperdrive does not collect or access the following:

  • Camera or microphone data. Hyperdrive does not request or use camera or microphone permissions.
  • macOS Contacts — write access. Hyperdrive reads macOS Contacts (via CNContactStore) for local autocomplete merging but never modifies, creates, or deletes entries in the system address book. Contact data read from macOS Contacts is cached locally and never transmitted to any server.
  • Location services. Hyperdrive does not request or use GPS, Wi-Fi-based, or other precise location services.
  • Bluetooth. Hyperdrive does not use Bluetooth.
  • Accessibility permissions. Hyperdrive does not request the macOS Accessibility permission and does not use accessibility APIs to monitor, inspect, or control other applications. Hyperdrive does use standard AppKit accessibility annotations (such as VoiceOver labels and roles) on its own user interface elements.
  • AppleScript or Automation. Hyperdrive does not use AppleScript or macOS automation frameworks.
  • Browsing history. Hyperdrive does not access or monitor your web browsing history.
  • Files on your disk (beyond its own data directories). Hyperdrive only reads and writes to its own application support directory, caches directory, and preferences file, plus files you explicitly select via file picker dialogs (such as email attachments).
  • Clipboard data. Hyperdrive does not monitor or log clipboard contents.

Hyperdrive does not embed any third-party advertising SDKs, analytics SDKs, crash reporting SDKs, or data broker integrations. Production dependencies include on-device machine learning libraries and models (which process data entirely locally) and the Sparkle framework (for software updates).


25. Your Responsibilities

Hyperdrive is a powerful tool that can access, modify, and generate email communications, calendar events, and Slack messages on your behalf. You are responsible for:

  • Account authorization. Ensuring that you have the legal authority to access the email accounts, Slack workspaces, and calendar accounts you connect to Hyperdrive.
  • Email tracking compliance. Ensuring that your use of email tracking features complies with all applicable laws and regulations, including laws that may require recipient consent or disclosure. See Section 11.3 for details.
  • AI content review. Reviewing all AI-generated content (draft replies, rewrites, summaries, event extractions) before sending, accepting, or acting on it. AI-generated content may contain errors, inaccuracies, or inappropriate language.
  • Workplace policies. Ensuring that your use of Hyperdrive complies with applicable workplace policies, employment agreements, confidentiality obligations, and professional conduct standards.
  • Scheduling feature accuracy. Verifying that scheduling links, availability windows, and calendar settings are accurate and appropriate before sharing them publicly.
  • Data sensitivity awareness. Being aware that certain Hyperdrive features transmit portions of your email and Slack content to third-party services (OpenAI, Google, Slack), and ensuring that such transmission is consistent with any data handling obligations, confidentiality agreements, or regulatory requirements that apply to your communications.
  • Custom AI endpoint. If you configure a custom AI endpoint and API key, you are solely responsible for evaluating the privacy, security, and data handling practices of that endpoint.

26. Server-Side Deferred Actions and Expanded Trust Boundary

Hyperdrive provides server-side deferred actions that operate when your Mac is offline or asleep. These features require storing certain data on Hyperdrive servers as described below.

26.1 Server-Side Capabilities

The following features are deployed and active when you enable cross-device sync:

  • Server-side snooze restoration: Server stores snooze records (thread identifier, account identifier, return date, labels to restore) and restores the INBOX label via Gmail API when the snooze fires. This requires the server to call the Gmail API on your behalf.
  • Server-side scheduled email sending: Server stores full MIME-encoded email content and sends the email via Gmail API at the scheduled time, even when your Mac is offline or closed. The server stores the complete email body until it is sent, and deletes it after successful delivery.
  • Server-side reminders: Server stores reminder records (thread identifier, account identifier, remind-at date, condition such as "if no reply," and a message count snapshot) and fires them by restoring the thread to your inbox via Gmail API (adding INBOX and UNREAD labels), following the same mechanism as snooze restoration. For "if no reply" reminders, the server queries the thread's current message count to determine whether a reply was received, and skips the reminder if so.
  • Server-side Gmail OAuth token storage: To perform snooze restoration, reminder firing, and scheduled sending, the server stores your Gmail OAuth refresh tokens. These tokens grant the server read and write access to your Gmail account (within the scopes you originally authorized). Tokens are encrypted at rest using AES-256-GCM with per-account random salts. However, the server has decryption capability at runtime — the encryption protects against database-level breaches but not against server-side application compromise.
  • Deferred action execution: All deferred actions (snoozes, reminders, scheduled sends) are executed directly by the server via the Gmail API. Push notifications are not currently used for deferred action delivery.

26.2 Data Stored Server-Side

When these features are enabled:

  • Hyperdrive servers store Gmail OAuth credentials that allow the server to read, modify, and send email on your behalf within the scopes you authorized.
  • Email activity patterns (snooze times, send-later schedules, reminder patterns) are stored on the server.
  • Full email content of scheduled messages is stored server-side until sent.

These features require explicit user consent before activation. You may revoke access at any time by deauthorizing Hyperdrive from your Google Account security settings.


27. Changes to This Privacy Policy

Hyperdrive reserves the right to update, modify, or replace this Privacy Policy from time to time to reflect changes in Hyperdrive features, legal requirements, regulatory guidance, or operational practices.

Notification of changes: If changes are material (such as new categories of data collection, new third-party data sharing, or changes to user rights), Hyperdrive will provide notice through one or more of the following means:

  • An in-application notification or banner within Hyperdrive.
  • A notice on the Hyperdrive website.
  • An email to the address associated with your license or account (for material changes).
  • Updating the "Last Updated" date at the top of this Privacy Policy.

Effective date: Changes will be effective as of the "Last Updated" date shown at the top of the revised Privacy Policy.

Acceptance: Continued use of Hyperdrive after an update to this Privacy Policy constitutes your acceptance of the updated Privacy Policy. If you do not agree with the changes, you must discontinue use of the Service.

We encourage you to review this Privacy Policy periodically to stay informed about our practices.


28. Contact Information

If you have questions, concerns, or requests regarding this Privacy Policy, your personal information, or the data practices described herein, please contact:

Hyperdrive One LLC 1431 Porter Rd, Nashville, TN 37206, United States Email: support@hyperdriveone.com

Hyperdrive One LLC is the data controller for the personal data processed through the Service.

When contacting Hyperdrive:

  • Identity verification: You may be asked to verify your identity before information is provided, modified, or deleted. This is to protect your privacy and prevent unauthorized access to your data.
  • Response time: Hyperdrive aims to respond to all inquiries within 30 days. For GDPR requests, we will respond within 30 days (extendable by up to 60 additional days for complex requests). For CCPA requests, we will respond within 45 days (extendable by up to 45 additional days).
  • Data protection inquiries: If your inquiry relates to GDPR, CCPA, or other data protection rights, please specify the right you are exercising and provide sufficient information for us to verify your identity and locate your data.

If you are located in the EEA and are not satisfied with our response to a data protection inquiry, you have the right to lodge a complaint with your local supervisory authority. A list of EEA supervisory authorities is available at the European Data Protection Board website (https://edpb.europa.eu/).

© 2026 Hyperdrive. All rights reserved.
Features Pricing Changelog Docs Privacy Terms Support

This site uses Cloudflare Web Analytics for anonymous usage statistics. No cookies are used for tracking.

Coming Soon

Hyperdrive is launching soon. Join the waitlist to get early access.