Subprocessor Information
Retrium Next (Beta)
Last Updated: August 10, 2026
To provide Retrium Next (the “Beta Service”), Retrium, Inc. engages certain third-party companies as subprocessors to process personal data on our behalf—for example, to host and store data, provide analytics, and deliver customer support and communications. This page lists the subprocessors we currently use for the Beta Service. We require our subprocessors to protect personal data consistent with our obligations and to use it only as needed to provide services to us.
Beta notice. Retrium Next is a beta service. This subprocessor list may change frequently as the product evolves, and we do not make data retention, backup, or uptime commitments during the beta. This list is provided for transparency and does not create any contractual commitment regarding a specific set of subprocessors.
Current Subprocessors
| Subprocessor | Purpose / Function | Country of Establishment | Processing Location |
|---|---|---|---|
| Amazon Web Services (AWS) | Cloud hosting and data storage | USA | USA |
| OpenRouter, Inc. | AI inference routing to Google Gemini (zero data retention; no training) | USA | USA |
| Google (Gemini API) | AI model provider for summaries/themes (zero data retention; no training) | USA | USA |
| MongoDB Atlas | Database hosting and data storage | USA | USA |
| Redis Labs | In-memory data store / caching | USA | USA |
| Functional Software, Inc. (Sentry) | Application error monitoring and performance/crash reporting | USA | USA |
| Statsig, Inc. | Feature flags, experimentation, and product analytics | USA | USA |
| Google Analytics | Website and product analytics | USA | USA |
| Google Ads | Marketing and advertising | USA | USA |
| Marketing and advertising | USA | USA | |
| Meta | Marketing and advertising | USA | USA |
| X (formerly Twitter) | Marketing and advertising | USA | USA |
| Marketing and advertising | USA | USA |
AI Processing
AI features in Retrium Next perform inference through OpenRouter using Google’s Gemini models as the sole model provider; requests are processed only on Google’s infrastructure. Every request is configured for zero data retention with model training explicitly disabled, so neither OpenRouter nor Google retains prompts or outputs or uses them to train models. Requests are tagged only with pseudonymous internal identifiers, never names, email addresses, or other direct identifiers, so the provider cannot link retrospective content to a specific individual unless a user includes personal information within the content itself.
Changes and Notifications
We may add or replace subprocessors as the Beta Service develops. When we make material changes to this list, we will update the “Last Updated” date above. You can request to receive notifications of changes to Retrium’s subprocessors, or ask questions about this list, by contacting us at next@retrium.com.