Privacy Policy

Last updated 2026-09-02

1. Who we are

Saido is operated by Dmytro Serhiiev, an individual entrepreneur registered in Ukraine. He is the owner and the data controller for this website and the Saido iOS app; our full registered address is available on request. Contact: [email protected] — that address reaches us for any privacy question or request. You can also verify who is behind Saido on LinkedIn.

2. The short version

Audio and transcripts are processed on your device.
The iPhone app has no account, so we never hold your notes.
The website has an optional account for the free transcript tool. It holds your email and the videos you opened — nothing from the app.
Optional AI features send text (never audio) to our API.
Summaries and translations on the website are generated by a model; we count how many you use, not what they say.
We never receive or store your card details; payments are processed by our payment provider.
Website analytics only, and in Europe only after you accept them.
No advertising SDKs, no sale of personal data, no training on your content.

3. Data processed in the app

Data	How it’s handled
Audio recordings	Stored locally; never uploaded.
Transcripts & notes	Local; optional iCloud via your Apple ID under Apple’s terms.
Text sent for AI structuring/chat	Transmitted to our server, then to Google’s Gemini API; not stored beyond the request except transient logs kept ≤ 30 days. The provider is bound by zero-retention / no-training terms.
Reminders	Written to Apple Reminders with your permission.
Purchase data	Handled by Apple; we receive anonymous receipt validation and never see your payment details.
Diagnostics	Crash reports, opt-in via iOS settings; no third-party analytics SDK at launch.

4. Data processed on the website

Server logs (IP, user agent, kept ≤ 30 days, hosting provider Cloudflare).
Waitlist email (stored until the launch mailing plus 90 days, or until you ask us to delete it).
YouTube tool: the link you paste goes to an endpoint on this site, which asks our API server (Hetzner, EU) for the captions YouTube already publishes for that video. The finished transcript is cached on Cloudflare KV (EU) for up to 30 days, keyed by video ID and caption language, so anyone opening the same video gets it instantly. That cache holds public captions and nothing about you. Your IP address is used only to rate-limit the tool, in hashed form, for one hour.
Daily limit for guests: to count the 2 transcripts a day a guest gets, we set a cookie with a random identifier and count, against a hashed form of it, which video IDs were opened today. Both the counter and the list expire within two days. The identifier is not linked to an account and is not used for anything else.
Optional account: the tool works without one. If you sign in, we store on Cloudflare D1 (EU) your email address, the dates you signed in, the country and browser string of each sign-in, and the list of videos you opened while signed in (video ID, title, channel, how many times). There is no password — signing in means clicking a one-time link we email you, and the link is stored only as a hash and expires in 15 minutes. Kept until you delete the account.
Paid plans: we never receive or store your card details. Payment is taken by our payment provider, which acts as the seller of record for website purchases and sends us only what we need to give you the plan and to keep our books — an order identifier, the plan, the amount, the date and the billing country.
AI features on the transcript page: a summary, and a question you ask about a video, send the transcript text — public captions, never audio, never anything from the iPhone app — to Google’s Gemini API, which is bound by zero-retention and no-training terms. A question you type is sent with it. Summaries and answers are cached for up to 30 days, keyed by video, language and (for answers) a hash of the question, the same way the transcript is, and hold nothing about you. Your conversation with a video is kept in your own browser and never stored on our servers.
AI actions: to keep to your plan’s limit we count how many AI actions your account has used today. We store your account ID, the date and the number — not the text we sent and not what came back. We also count, against a hashed identifier, how many questions you asked about a given video today, so that the second and later questions about it are free.
Share links: if you create a link to share a transcript, we store on Cloudflare D1 (EU) the video, the caption language, which parts you chose to include, when the link was made, when it expires, how many times it has been opened, and a name only if you typed one yourself. The link’s own address is stored only as a hash, so nobody — including us — can recover a working link from the database. Your questions and answers about a video are never shareable and are never part of a link. Revoking a link stops it working immediately; revoked links are kept as a record until you delete your account.
Product analytics: we record events about how the tool is used — page opened, transcript requested, whether it succeeded, whether it came from cache, how long it took, export and copy actions, and the sign-in steps. Each event carries the country, and either your account ID (if you are signed in) or a pseudonymous daily identifier derived from a salted hash of your IP address and browser string. That daily identifier changes every day and cannot be used to follow you across days. Events are deleted after 180 days. If you delete your account, your account ID is stripped from past events and they stay as anonymous counts.
Third-party analytics: we load Google Analytics 4 and Microsoft Clarity through Google Tag Manager to see which pages work and where people get stuck. Neither is loaded on the transcript tool’s account pages or on a shared transcript link’s page — those show personal content, and the address of a shared link is itself the key to it. Clarity also records anonymised session replays and heatmaps, with its Balanced masking so text you type and content you view are hidden. Google Analytics keeps this data for 14 months. Advertising storage is switched off everywhere, always — we run no ads and build no ad audiences.
English Dictionary analytics: the dictionary uses a separate Google Analytics 4 web stream in the same Saido property, including its signed-in app and shared-lesson screens. It loads only after your browser’s saved analytics choice is “allow”; a new dictionary visitor is asked to opt in regardless of location. We send screen names, sign-in request/delivery steps, onboarding actions, confirmed assistant connection, successful web imports and sharing actions, and practice counts and self-assessments. We do not send emails, account IDs, card text, lesson names, support messages, file names, tokens, shared-link identifiers, query strings, or URL fragments. Each page URL is reduced to the public landing, Ukrainian landing, or app address. Automatic form/link/history tracking is disabled for this stream, and Microsoft Clarity is not loaded in the dictionary. You can change this choice using “Analytics preferences” at the bottom of dictionary pages. Material creation by your connected assistant is not sent to Google from our server. Standard Google Analytics browser and network metadata, such as approximate geography and device information, are processed after consent.
Your choice about analytics: if you are in the EEA, the UK or Switzerland, both tools stay off until you accept them in the banner at the bottom of the page. Everywhere else they start on and the banner lets you turn them off. Your answer is remembered in your browser under saido.consent, not on our servers, and you can change it by clearing your site data.
Cookies: sd_gid is strictly necessary and set for everyone — it is the random identifier behind the daily guest limit described above, and holds nothing else. sd_sess and sd_who are strictly necessary and set only after you sign in — the first is the signed session that keeps you signed in for 30 days, the second holds your email so the page can show it without asking the server. Both are cleared when you sign out. Google Analytics and Microsoft Clarity set their own analytics cookies, but only once analytics is allowed as described above — that is what the consent banner is for.

Contract (providing the service), legitimate interest (security, abuse prevention), consent (waitlist email, diagnostics, and third-party analytics where consent is required).

Processors: Cloudflare (hosting, CDN, transcript cache, account and analytics database), Hetzner Online GmbH (API server, EU), Resend (sending sign-in emails), Google (Tag Manager, Analytics 4, and the Gemini API that structures notes), Microsoft (Clarity), our payment provider (taking payment, invoicing and billing correspondence for website purchases), Apple (App Store, iCloud). We do not sell your data and do not use ad networks.

7. International transfers

EU servers for the API, the transcript cache and the account database. Google (Tag Manager, Analytics, Gemini) and Microsoft (Clarity) may process data in the US under SCCs / the EU-US Data Privacy Framework.

8. Retention

Device data: until you delete it. Server transient logs: ≤ 30 days. Waitlist, account data, and our own analytics events: see §4. Google Analytics data: 14 months. Sign-in links expire after 15 minutes; sessions after 30 days. Payment metadata (the order identifier, plan, amount, date and billing country described in §4): kept for 3 years (1095 days) from the payment, as accounting and tax law requires, and then deleted. Daily AI action counters: 13 months, so a day can be compared with the same period a year earlier; deleted with your account. Card data we never hold at all.

9. Your rights

You have the right to access, rectify, erase, port, and object to processing of your data, and to complain to your supervisory authority. If you have an account, two of these need no email to us: Download my data and Delete account are on your account page and take effect immediately. California residents have CCPA rights; we do not sell or share personal information.

10. Recording other people

You are responsible for complying with consent laws where you record. Saido shows a reminder before meeting recordings.

11. Children

Saido is not directed to individuals under 16.

12. Changes

We post updates here with a new date. Material changes are notified in-app.

Questions? [email protected]