Draft — for counsel review before launch
Version 0.1 (draft), 3 November 2026. Highlighted text marks facts still to be confirmed.
Subprocessors
Nanner Tech LLC uses these subprocessors to run Rhumb. The privacy policy and the data processing addendum say how they fit in.
Current subprocessors
| Subprocessor | Purpose | Data it receives | Location |
|---|---|---|---|
| Google Cloud and Firebase | Hosting the website and API, the database, secrets and catalog storage; Firebase Authentication for sign-in | All service data, including prompts while they are answered, and sign-in credentials | United States (API and database in the us-central1 region) |
| OpenAI | The decision model that reads prompts sent to /v1/plan and the sandbox's prompt mode | The first 4,000 tokens of each prompt, its tags and its attachment types | United States. [To be confirmed by counsel: zero data retention status and EU regional processing] |
| Stripe | Payments, subscriptions, invoices and the billing portal | Account email, payment details entered on Stripe's pages, subscription and invoice records | United States |
| MailerSend | Sending account, quota and billing emails | Email address and message content | [To be confirmed by counsel: MailerSend's processing location] |
| Cloudflare | Turnstile challenge on the sandbox, sign-up and password reset | IP address and browser signals during the challenge | [To be confirmed by counsel: Cloudflare's processing location for Turnstile] |
| Sentry | Error reports from the API | Error details and the request path, without request bodies, query strings, cookies, secret headers or user details | [To be confirmed by counsel: Sentry's data region for our project] |
Changes to this list
Before we add or replace a subprocessor, we email every account owner and update this page, so API customers can object under the data processing addendum. [To be confirmed by counsel: notice period before a new subprocessor starts]