Subprocessors

Last updated: 6 September 2026

This page lists the third-party companies (“subprocessors”) that process data on Auserene’s behalf to operate the app. We send each one the minimum data it needs, and we choose providers for our most sensitive data based on their data-handling terms. For how this fits into our overall data practices, see our Privacy Policy.

A core principle: your raw journal entries and conversations are sent only to our language-model and embedding providers, which are contractually prohibited from training on your content. Our text-to-speech providers receive only the short generated meditation script — never your raw journal text. Our analytics, subscription, and sign-in providers never receive any note, voice note, mood entry, or conversation content.


Language model & embedding providers

These receive your journal/conversation content (decrypted in our secure server environment only to build each request) to generate the companion’s responses, reflections, and the searchable memory that gives the companion continuity. All are contractually prohibited from training on your content and delete inputs on short retention schedules (approximately 30 days or sooner).

ProviderPurposeData received
AnthropicGenerates the companion’s responses and session reflectionsSession conversations and derived context
OpenAIGenerates supporting text and creates embeddings used for memory/recallSession and journal content; derived context
Fireworks AIConversational and batch text generationSame content as above, when this route is used — under zero-data-retention terms: not stored after the response is generated

Speech-to-text providers

These receive the audio of a voice note, return the transcript, and keep nothing. Both process audio under zero-data-retention terms, and we do not store the audio either. They never receive your written journal or conversation history.

ProviderPurposeData received
GroqTranscribes voice notesVoice-note audio, transiently; zero data retention
Together AITranscribes voice notesVoice-note audio, transiently; zero data retention

Text-to-speech providers

These receive only the short, generated meditation script to turn into audio. They never receive your raw journal text or conversations.

ProviderPurposeData received
ElevenLabsVoices generated meditation scriptsShort generated script text only
Inworld AIVoices generated meditation scriptsShort generated script text only

Analytics, subscriptions & sign-in providers

These receive an internal user identifier and event or device data. None of them receives any note, mood, or conversation content.

ProviderPurposeData received
PostHogProduct analytics — optional, off via Settings → Share anonymous usageAction events only (note saved, session started, meditation completed), internal user id, app version, device type. No advertising identifier, no cross-app tracking
SuperwallSubscription and paywall serviceInternal user id, purchase and subscription events from the App Store, device attributes for paywall display (model, OS and app version, locale, screen size)
AppleApp Store, StoreKit, Sign in with ApplePurchases and subscriptions under Apple’s terms; sign-in identifier if you use Sign in with Apple
Google (Google LLC)Google Sign-InSign-in identifier and Google account email, if you use Google Sign-In

Infrastructure & security providers

ProviderPurposeData received
Google Firebase (Google LLC)Authentication (email, phone, Apple, Google), database, file storage, and server functions — our core hostingAll stored app data (encrypted at rest where sensitive)
Google Cloud KMS (Google LLC)Holds the master key that protects each user’s encryption keyEncryption key material only — no user content

Changes to this list

We may add, remove, or change subprocessors as the app evolves. When we do, we update this page and the date above. For material changes, we’ll also note it in the app where appropriate.

Questions: hpatsvnit@gmail.com