Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.hyprcore.ai/llms.txt

Use this file to discover all available pages before exploring further.

Privacy is Hyprcore’s main differentiator. Here’s specifically what that means.

​
What’s on your Mac, always

  • Audio recordings (dictation WAVs, meeting audio/video files).
  • Transcripts and notes.
  • Vault structure (folders, page titles, page bodies).
  • Speech models. Whisper, Parakeet, and any custom models live entirely on disk.
  • API keys. Stored in macOS Keychain. They never reach Hyprcore servers.
  • Dictation history. Always local; never syncs.
By default that’s where it stays. Sign-in proves who you are; it doesn’t share what you’ve recorded. Cloud sync, cloud STT, and cloud AI features are all opt-in—when they’re off, Hyprcore only contacts the network for auth, auto-updates, and one-time model downloads.

​
What can reach the cloud, only with your action

​
When you sign in and turn on Cloud Sync

  • Vault metadata (folder structure, page titles).
  • Transcripts and notes (encrypted blobs—our servers can’t read them).
  • Action items.
  • Audio and video files, only if “Sync audio” is enabled. Off by default.
Synced data lives on AWS S3, encrypted in transit and at rest. The sync service uses authenticated, scoped credentials per device.

​
When you use Hyprcore Cloud AI features

For each AI request you make—a meeting summary, a session chat question, a cloud STT transcription—the relevant content travels to the third-party LLM or STT provider that fulfills the request:
  • AI summaries / dictation cleanup / chat: Anthropic, OpenAI, Google, or another LLM provider via OpenRouter (when using Hyprcore Cloud) or directly (when using your own API key).
  • Cloud STT: Deepgram, Groq, or OpenAI.
  • Cloud embeddings (semantic search): OpenAI or another provider via OpenRouter.
These providers process your content to fulfill the immediate request and don’t retain it for training. Hyprcore Cloud uses zero-data-retention versions of every provider where available.

​
When you use BYO API keys

Same data flow, except Hyprcore is invisible—your API key calls the provider directly from your Mac. We don’t see the request or the response. You’re billed by the provider, not by us.

​
What’s never collected

  • Audio or video. Never sent to Hyprcore’s telemetry or logging systems.
  • Transcript content. Never sent to telemetry. Sent to LLM providers only when you ask for an AI feature.
  • Personal identifiers beyond what’s needed for billing (your account email, billing details—handled by Stripe).
  • Browsing or app usage outside Hyprcore.

​
Telemetry

Hyprcore collects crash reports and (optionally) anonymized feature usage. Default is Errors only—just crashes. Default never includes content or audio. Configure or disable in Settings → Telemetry. See Telemetry for the full breakdown.

​
Storage and retention

DataWhereRetention
Local meetings, notes, historyYour MacForever, or per your retention setting
Cloud-synced meetings, notesAWS S3While your account is active; 30 days after cancellation
Cloud-synced audio/videoAWS S3 (only if Sync Audio on)Same as above
Telemetry errorsSentry90 days
Telemetry logs (Full mode)Axiom30 days
Stripe billing dataStripePer Stripe’s terms

​
Your rights

If you’re in the EU, UK, California, or any jurisdiction with comparable rights:
  • Access. Request all data Hyprcore stores about you. Email privacy@hyprcore.ai.
  • Deletion. Request deletion of your account and all associated data. Email privacy@hyprcore.ai. We delete cloud data within 30 days; local data is yours to delete via rm -rf ~/Library/Application\ Support/ai.hyprcore.desktop/.
  • Portability. Export meetings as Markdown or JSON from inside the app. For a full data export, email privacy@hyprcore.ai.

​
Subprocessors

Hyprcore Cloud uses these subprocessors:
  • AWS — storage (S3) and serverless compute.
  • Supabase — auth and managed Postgres.
  • Stripe — billing.
  • Sentry — crash reporting.
  • Axiom — log shipping (Full telemetry mode only).
  • OpenRouter — multi-provider LLM gateway used by Hyprcore Cloud AI features.
  • Anthropic, OpenAI, Google, Deepgram, Groq — model providers (when you use Hyprcore Cloud and a request routes to them).
For the current list with addresses, see hyprcore.ai/privacy.

​
Contact

For privacy questions or requests: privacy@hyprcore.ai. For everything else: support@hyprcore.ai.