Subprocessors

Company
Strix AI, Inc.
Updated
November 20, 2026

Our Privacy Policy explains how we use personal information.

These are the outside services that receive personal data when STRIX runs the Services, where each one stores that data, and whether it uses the data to train its own models. Our Privacy Policy explains how we use personal information.


Services that process personal data

Service Data it receives Where data is stored Data processing agreement Trains on our data
Amazon Web Services Hosting, compute, file storage, queues, cache, logs, and content delivery. All Service data, including account and client profiles, health notes, workout logs, uploaded files, signed forms, and request logs. United States (Oregon, us-west-2) In standard terms No
PlanetScale Primary Postgres database, reached over a private AWS link. All Service data held in the database, including accounts, client profiles, health notes, workout logs, messages, and billing identifiers. United States (Oregon, us-west-2) In standard terms No
Cloudflare DNS, the network proxy in front of the web app, and Turnstile bot checks on sign-up. Web app request traffic, IP addresses, and browser signals. Global network, served from the data center nearest the user In standard terms Turnstile signals train its bot detection
OpenAI AI workout plan generation and editing, plan and document imports, and the Daily Pulse briefing. Client details used to build a plan: age, sex, height, weight, goals, medical conditions, injuries, and coach notes. Also plan documents coaches upload. United States Requests are not pinned to a region, so OpenAI may process them outside the United States. Signed No
Anthropic The in-app help assistant, plan editing, report and automation drafts, client filters, and Dumby Scribe intake drafts. Coach, team, and client names, client status, goals, and injuries, and plan content. Also transcripts of Dumby Scribe recordings, which can include health information. United States Requests are not pinned to a region, so Anthropic may process them outside the United States. In standard terms No
AssemblyAI Speech-to-text with speaker labels for Dumby Scribe recordings of coach and client conversations. Audio of the conversation and its transcript, which can include a client's goals, injuries, medical conditions, and other health information. United States AssemblyAI downloads each recording from a short-lived link. Strix deletes the transcript from AssemblyAI as soon as it is saved in Strix. In standard terms Only if not opted out
Voyage AI Search embeddings and ranking for the exercise library. Exercise text and the search queries coaches type, which can mention a client. United States In standard terms No (opted out)
Stripe Strix subscriptions, coach-to-client billing, sales tax, and card collection. Names, email addresses, billing addresses, card details (collected by Stripe, never stored by Strix), invoices, transactions, and device data. United States In standard terms Yes, for its fraud models
Twilio SMS sign-in codes, form links, and text messages coaches send. Phone numbers, message content, verification codes, and delivery status. United States In standard terms Yes, for abuse prevention
Resend Account, billing, and form email, and email coaches send to clients. Names, email addresses, and email content. United States In standard terms No
Mux Video upload, storage, and playback. Videos coaches upload, and viewers' IP addresses, devices, and playback events. United States Mux Data shortens viewer IP addresses in Germany before sending playback data to the United States. In standard terms Anonymized data only
Ably Live updates for schedules, check-ins, and plans. Update contents, which can include client names, user identifiers, and IP addresses. Global network, served from the data center nearest the user In standard terms No
Meilisearch Search across clients, coaches, exercises, and plans. Names, email addresses, and phone numbers of users, coaches, and clients, with their team and location. To confirm The search host is kept as a secret. Check whether it is Meilisearch Cloud and which region it runs in. None published No
Sentry Error tracking, performance tracing, and session replays with text and inputs masked. Error reports, request details, user identifiers, masked session replays, and the name and email on in-app feedback. United States Accepted Non-identifying usage data only
Statsig (Amplitude) Feature flags and staged releases. User identifier, name, email, role, team, and location. United States In standard terms Not stated
Amplitude Product analytics for the web and mobile apps. User identifier, name, email, role, team, pages viewed, clicks, device data, and IP address. United States In standard terms Not stated

Processing outside the United States

Strix stores Service data in the United States. The application, database, and uploaded files run in Amazon Web Services and PlanetScale in Oregon. No Strix team member or contractor reaches user data from outside the United States.

Some services can process data outside the United States:

  • Cloudflare and Ably route each request through the data center nearest the user.
  • Anthropic and OpenAI may run AI requests in other regions because Strix does not pin them to the United States.
  • Mux Data shortens viewer IP addresses in Germany.
  • Stripe, Twilio, and Sentry store data in the United States, and their terms allow processing in other countries.
  • Meilisearch is a French company, and the search index region is not yet confirmed.

Sign-in providers and services users choose

These services receive data only when a user or team turns them on, or when a user's browser loads content from them.

Google
Sign in with Google. If a user adds their Strix schedule to Google Calendar, Google fetches that schedule.
Apple
Sign in with Apple. Workouts are written to Apple Health on the user's own device; Strix reads nothing back.
YouTube and Vimeo
Videos coaches link as resources load from these sites, which receive the viewer's IP address and cookies.
UI Avatars and Bunny Fonts
Default profile images and one sign-in page font load from these services, which receive the viewer's IP address and, for avatars, initials.

Services we do not use

  • GoHighLevel
  • Terra
  • Direct wearable connections (Garmin, WHOOP, Oura, Google Health, Polar). These are built but not released, and each will be listed before it goes live.
  • Push notification services
  • Video call providers
  • Advertising or marketing trackers. The website loads no analytics or tag scripts.

Open items before publishing

  1. Ask Meilisearch for a data processing agreement, since none is published, and confirm the search project's region.
  2. Decide whether to pin Anthropic and OpenAI requests to the United States.
  3. Confirm the Statsig agreement now that Amplitude runs Statsig.
  4. Turn on AssemblyAI's model training opt-out in its Data Controls, then change its training answer to No.