Service providers and subprocessors
The companies we rely on to run Jinova, and what each one handles.
Running Jinova means relying on other companies — for hosting, storage, payments, email, and the compute that processes your files. This page lists them and what each one handles.
The legal role a provider plays can differ depending on the processing involved, so we describe what each one actually does rather than applying a single label across the list.
For the content of your files, which we process on your behalf under our Data Processing Terms, the providers marked below as receiving your uploaded files act as our subprocessors.
Who we use
| Provider | What it does for us | What can reach it | Your uploaded files? |
|---|---|---|---|
| Neon | Hosts our database | Your email, account and sign-in records, job details and settings, balance and transaction history, payment references | No |
| Cloudflare (R2) | Stores your files | Your uploaded files, the results, and execution records | Yes |
| Vercel | Hosts and runs the console | Requests to the site, IP addresses, operational logs, and the limited parts of your files described under How your files move below | Partly — uploads and downloads go straight between your browser and storage, but our service reads limited parts of a file to check it |
| Stripe | Takes card payments | Your email, the amount, and a reference to your account — plus the reference for one of our ads, if you arrived from one and allowed advertising measurement. Card details are entered on Stripe’s own pages and we never see them | No |
| RunPod | Runs AI jobs on GPUs | A time-limited link to your files, your job settings, and a link to write the result back | Yes |
| Modal | Runs AI jobs on GPUs | The same as RunPod, for the tools it runs. Transcription is not sent to it | Yes |
| Resend | Sends sign-in emails | Your email address and the sign-in link | No |
| Optional “Continue with Google” sign-in | Only if you choose it: the sign-in exchange with Google | No | |
| Microsoft | Optional “Continue with Microsoft” sign-in | Only if you choose it: the sign-in exchange with Microsoft | No |
| Google Analytics | Usage analytics — only if you allow it | Pages visited and a limited set of usage events. Never your files, results or amounts | No |
Independent controllers
The company below is not our subprocessor. It receives information from us for its own purposes, as an independent controller, under its own advertising terms and privacy policy.
| Company | Why it receives anything | What we send it | Your uploaded files? |
|---|---|---|---|
| OpenAI | Measures, and reports to us on, whether our ads in ChatGPT lead to sign-ups and credit purchases — only if you allow advertising measurement | Only if you arrived from one of our ads: the reference for that ad; that you signed up or bought credit, with the amount and currency of a purchase; and a fixed page address our servers set for that event. Never the address actually in your browser or its query string, a run identifier, a Stripe payment reference, or your name, email, files or results | No |
How your files move
Your original upload travels directly between your browser and our storage — it is not routed through the service that runs the console. Results travel the same way: the machine that did the work writes them straight into our storage, and a download goes straight from storage to you.
Our service does read limited parts of your files, for three reasons. Before we price a recording, we read enough of it to measure how long it is. Before we treat a result as delivered — and before you are charged for it — we read the result to check it is the thing you asked for. And when you open a transcript in the console, we read it so we can show it to you. Nothing read for those checks is kept.
Transferring files between our storage and the machine that processes them is done with links we create for one file at a time, which stop working shortly afterwards. The machine fetches your recording and writes its results using those links.
Transcripts are not put into our analytics or operational records. What we keep about a transcription is counts and timings — how long the recording was, how many segments it produced, which language was detected — not the words.
Changes
If we plan to add or replace a provider that receives your uploaded files, we will update this page at least 30 days before it starts receiving them — unless a change is needed urgently to keep the service secure or available, in which case we will update it as soon as we reasonably can. Our Data Processing Terms explain how to object.
Questions about any of this go to our support page, under Privacy or data request.