API privacy policy

What we process

We process the audio in each file you submit to produce a transcript, word timestamps and speaker labels. We also store your account email, password hash, recovery-code hash, API-key hashes, sessions, payment references, credit balance and usage records, including when a finished result is downloaded. Your email is a sign-in identifier and is not verified.

We review account, job, usage and payment records, including the server records of Kalemio app installs, per account or install in a private, owner-only dashboard to see where people get stuck and to fix failures. It shows pseudonymous IDs, not email addresses, and uses no third-party trackers.

Data flow

This follows one API upload through the service as it runs today. It describes current operation; it is not a contract.

1. Upload. Your request reaches the Kalemio API on Cloudflare, which measures the recording and stores it in Cloudflare R2. Once our transcription gateway has received the recording, that copy is deleted.
2. Gateway. The gateway, also on Cloudflare, keeps the recording in R2 while the job runs and deletes it when the job finishes. An hourly cleanup removes any recording older than 24 hours.
3. Transcription. The recording is transcribed on a server we operate or, for paid plans, on RunPod, a cloud compute provider that runs our transcription engine as a subprocessor. Free plan jobs run only on our own server. On RunPod, the gateway sends the recording with each request and our engine returns the transcript; the engine keeps no job records. Our own transcription model does the work; the API does not send recordings or transcripts to other AI services.
4. Result. The transcript is stored in Cloudflare R2 for you to download. The gateway deletes its copy shortly after the job completes, retrying if that fails, and removes anything left after 7 days. Your downloadable result is deleted 24 hours after submission; after that, downloads return 410.
5. Billing. With usage-based billing, Stripe receives each job’s ID and billed minutes, not audio or text.

What stays after a job. The job record: its ID, your Idempotency-Key, a SHA-256 fingerprint of the uploaded file, its duration, speaker setting, status, billing method and times, and whether the result was downloaded. These records keep billing correct and stop a retried upload from being charged twice. Database backups keep deleted rows for up to 30 days.

Where. We don’t pin storage or processing to a country or region, so your data may be processed outside your country.

Agreements. We don’t publish a data processing agreement for the API. If your organization needs one, or other contract terms, email Support email and tell us what you need.

Audio and result retention

Uploaded media is not used to train models. Working audio and upstream job files are deleted after processing. Downloadable results expire 24 hours after submission. Cleanup retries during service outages, so physical deletion can be delayed by an outage. Download any results you need before they expire.

Service providers

Cloudflare hosts the website, API gateway, account database and temporary storage. Our own transcription server and RunPod, our compute provider for paid-plan jobs, process uploaded media to return results. Stripe processes payments and card details; card details are not stored by Kalemio. Data may be processed outside your country.

Account and payment records

Account, payment and usage records are retained for service operation, accounting and abuse prevention. Support correspondence is retained while needed to resolve your request. Previous waitlist records remain until they are no longer needed or you ask us to remove them.

Your choices

Revoke unused keys in your account. To request access, correction or deletion of personal information, email Support email. We verify requests before changing or deleting account data. Some payment, tax, security or legal records may need to be retained; we will explain any exception. See account deletion.

Website and security

We do not sell personal information or share recordings with advertising networks. The website uses privacy-limited, cookieless Google Analytics with storage denied to measure page paths and general product actions. It does not receive audio, transcripts, filenames, account details, payment details or API keys, and we do not send a persistent user identifier. You can turn it off with “Analytics settings” in the footer. We do not use personalized advertising trackers. Sign-in uses a session cookie. Requests use HTTPS, and the service uses authentication and rate limits to restrict access.