Subprocessors
Draft pending review. This page is published for transparency while the wording is still being reviewed, so it can change before launch.
These are the vendors that process data on our behalf when you run a conversion. What each one receives and how long it keeps it is described in the Data Use policy.
| Vendor | Role | Processing location |
|---|---|---|
| OpenAI | Generates components and runs the visual judge | United States |
| Google (Gemini) | Generates components and runs the visual judge | United States and other Google regions |
| E2B | Ephemeral sandboxes that install, build, render and screenshot generated code | United States |
| Cloudflare (R2) | Stores design images, screenshots, code bundles and permanent static previews | Cloudflare global network |
| Managed PostgreSQL host | Stores accounts, encrypted tokens, job and section records and reports | Region configured for the instance |
| Managed Redis host | Holds short-lived job status, queue messages and preview pointers | Region configured for the instance |
| Sentry | Receives error and crash reports from the API and the worker | United States or European Union, per project setting |
| Langfuse | Pipeline traces (timings and token usage) when tracing is switched on; design content only if that is enabled separately | Region configured for the project |
| GitHub | Sign-in, and pushing generated code to a repository when you ask for it | United States |
| Figma | Source of the designs, read with the access token you supply | United States |
Changes to this list
We will update this page before a new subprocessor starts handling customer data, and the date at the top shows the current version. If you want to be told about changes directly, subscribe to the project issue tracker, where changes are announced.