Privacy Policy

Last updated: October 31, 2026

Machina (“we”) is a personal knowledge base: you save links, text, and images, and Machina analyzes them with AI so you can search them and ask questions about them later. Doing that means handling things you wrote and things you chose to keep, so this policy is specific rather than general: it names the companies involved and lists what each feature actually sends them.

1. Who we are

Machina is an independent app operated from Israel by its developer, who is the data controller for the personal data described here. For any privacy question or request, including the rights described in section 10, contact support@mymachina.app. This policy covers the Machina iOS app, the Machina web app, and the browser extension.

2. What we collect

• Account information. When you sign in with Google or Apple we receive your name, email address, and profile photo from Firebase Authentication. Sign in with Apple lets you hide your real email address, and Machina works normally if you do.
• Content you save. The addresses you share, text extracted from those pages, titles and preview images, images and screenshots you share, and the notes, tags, categories, and collections you add yourself.
• Browser extension. The Machina browser extension reads a page only when you save it (with its button, its keyboard shortcut, or its right-click menu), and sends Machina only what that save needs: the page’s address and title; any text you selected on it; for an Instagram or TikTok post, the post’s preview (caption, cover image, and author) as the page shows it; and, when you ask for one, a screenshot of what you can see in the tab. It does not read, record, or send anything about pages you don’t save. It keeps a connection token, and the email address of the account it saves to, in that browser’s extension storage: the token is only ever sent to our server, and the email address is never sent anywhere. Saves it couldn’t send yet (while you’re offline, say) wait in the same storage until they go out or you discard them; saves from private windows are never stored. On mymachina.app it runs a small script so you can connect it in one click. The use of information received from the extension adheres to the Chrome Web Store User Data Policy, including the Limited Use requirements.
• Questions you ask. Your “Ask Machina” questions and the resulting chat history, so you can revisit past conversations.
• Preferences. Your in-app settings, including your device timezone, which is used to schedule reminders and digests at sensible local times.
• Notification token. If you turn on notifications on your iPhone, the push token Apple and Firebase issue for this app on that device, used only to deliver your reminders and digests. It is removed when you sign out or delete your account.
• Diagnostics. When something breaks, Machina records the error, with its message and stack trace, so we can fix it. These records live in your own workspace in our own database, they never include the content of your saves, and no third-party crash or analytics service is involved. Machina does not record which features you use.

What we do not collect: no third-party analytics or tracking SDKs, no advertising identifiers, no location, no contacts, and no browsing history beyond the pages you explicitly choose to save.

3. How we use your data, and on what basis

Everything we collect is used to run the product for you: analyzing saved content into summaries, categories, tags, and connections; semantic search; answering your questions with citations to your own saves; and sending the reminders and digests you switch on. We do not use your data for advertising and we do not sell it. Your content is not used to train AI models: not by us, and not by Google, as described in section 4.

Where data-protection law requires us to name a legal basis, ours are: performing the service you asked for (storing and organizing your saves, answering your questions); your consent, which you give on first run before any content is sent for AI analysis, and again when you enable notifications; and our legitimate interest in keeping the service secure, diagnosing failures, and preventing abuse. You can withdraw consent by deleting your account, which is described in section 9.

4. What we send to Google Gemini

Machina’s AI features are powered by the Google Gemini API. This is the part of Machina that involves another company, so here is the complete list, feature by feature:

• Saving a link. The text of the page we fetched, up to 30,000 characters, any text you shared or selected with it (for an Instagram or TikTok post, the caption your phone or browser showed), and your list of existing tags so the new card can reuse them. Usually only the site’s name goes with it. When the page can’t be read in full (a login wall, a post that only shows a preview), its address goes too, so the card can still be named, but never the part after a “?” or “#”, where links carry sign-in and tracking codes.
• Saving a post with photos, or a screenshot. The above, plus the image itself.
• Saving a YouTube video. The video’s address. Google watches the video on its own servers in order to summarize it.
• Writing a note. The text of the note.
• Search. A card’s summary text, converted into the numeric vector that makes semantic search and related-card links work. When you search, the words you typed, and the titles, the start of the summaries, and the tags of up to 20 matching cards, so they can be ranked by meaning. Private cards are never among them.
• Asking a question. Your question, the last few messages of that conversation, and up to 20 of your saved cards: their titles, summaries, details, and any notes you wrote on them.
• The weekly synthesis. The titles, summaries, tags, and categories of what you saved that week.
• Publishing a public page. Before a card, collection, or answer goes public, the text the page will show and its main images are checked by Gemini against our content rules (see section 8).

What is never sent: your name, email address, phone number, sign-in token, IP address, or any other account identifier. These requests are made by Machina’s servers rather than by your device, and every Machina user shares a single API key, so Google cannot tell one user’s requests from another’s, let alone connect them to you. What the content itself reveals is a separate matter: a note that names its author is still words in a request, and no technical measure changes that.

Machina uses the paid tier of the Gemini API. Under Google’s API terms for that tier, your prompts and responses are not used to improve or train Google’s models, and are not read by reviewers in order to improve them. Google does retain them for up to 55 days for the sole purpose of detecting abuse of the API, after which they are deleted. One exception sits inside that window and we would rather state it than let you discover it: content that Google’s automated systems flag as possibly breaking its usage policies can be assessed by authorized Google staff, to confirm or overturn the flag.

5. Private cards

A card you mark private (on its own, or by placing it in a private collection) is excluded from Ask, from digests, and from the weekly synthesis, and its tags are left out of the tag list described in section 4. Once a card is private, nothing about it is sent to Gemini again.

Two limits on that are worth stating plainly. First, a card is analyzed at the moment you save it (that analysis is what produces its summary and tags), so its content did reach Gemini then; marking it private afterwards stops all future processing but cannot undo the original request. Second, the optional PIN is a lock on the screen of this device, not encryption: private cards are stored the same way as everything else in your workspace.

6. Service providers

We use a deliberately short list of providers, each processing your data only to deliver the service:

• Google Gemini. AI analysis, summaries, embeddings, and answers, on the terms set out in section 4.
• Google Firebase and Google Cloud. Storage of your workspace, authentication, file storage, and the backend that runs Machina’s server-side logic.
• Vercel. Serves the web application, and receives standard web-server request logs, including IP addresses, in order to do so.
• RevenueCat. Manages Machina Pro subscriptions in the iOS app. It receives your account identifier (an opaque sign-in id, not your name or email), your App Store purchase receipts, and basic device information from the App Store, so that your subscription status can be checked and restored. Payment itself is handled by Apple; we never see your card details.
• Public post viewers. X (Twitter) and Instagram often refuse to show a public post to a server. To read one you saved, Machina’s server may request it through a third-party viewer service (fxtwitter, vxtwitter, or an Instagram viewer). They receive the post’s address, with any tracking parameters removed, and nothing else: not your account, not your other saves.

Each of these providers receives only the data listed here, uses it only to provide its service to Machina, and is bound by its own terms to protect it at least as well as this policy does.

Two small Google services also run on your device rather than on our servers: the web app loads reCAPTCHA to tell real browsers from bots, and the in-app link form fetches a site favicon from Google for the address you are saving. Both see your IP address and, for the favicon, the domain being saved, but never your account or your saved content.

7. Where your data is stored

Your workspace is stored in Google Firebase (Cloud Firestore, Cloud Storage, and Cloud Functions) in the United States (region us-central1). The web app is served by Vercel. All traffic between your device, our servers, and our providers uses HTTPS/TLS.

If you use Machina from outside the United States, your data is transferred there and processed there. Where such a transfer is restricted by the law of your country (for example within the European Economic Area, the United Kingdom, or Switzerland), it relies on the safeguards our providers offer for international transfers, including the European Commission’s Standard Contractual Clauses.

8. Public share pages

Sharing is off by default and nothing you save is public unless you choose to publish it. If you explicitly publish a card, a collection, or an answer from Ask Machina as a public Machina page, a snapshot of that content becomes visible to anyone holding the link, until you unpublish or delete it. Your name and email are never on the page. Deleting your account also removes every share page you published. Copies that other people or link-preview services made before that are outside our control.

Before a page goes live, its text and main images are checked automatically, and a page that looks like it breaks the content rules in our Terms is not published. Every public page has a “Report this page” link. We review reports within 24 hours, remove pages that break the rules, and can stop an account from publishing. We keep a record of that decision, tied to the account, for as long as it applies.

9. How long we keep your data, and how to delete it

We keep your data for as long as your account exists, because the product’s purpose is to still have your saves years from now. Diagnostic error records are an exception and are discarded automatically after 14 days.

You can delete your account from Settings at any time, without asking us. Doing so permanently removes your workspace (saved items, chats, and collections) along with your uploaded images and your sign-in record. You can also email us and we will delete the account for you. Deletion is not reversible, so export first if you want a copy.

Two small records outlive an account, and neither holds your content. So that deleting and re-creating an account does not restart the free trial, we keep a one-way hash of your email address with the date your first account was created; it cannot be turned back into your address and is used for nothing else. Abuse-prevention counters (how many requests an account or network address made in the last hour or day) expire on their own within a day.

10. Your rights

You can read, edit, and delete everything you have saved directly in the app, and export all of it (your cards and your notes on them, collections, and Ask conversations) from Settings, on the web or in the iPhone app, as a JSON backup plus a readable Markdown file, no request to us required.

Depending on where you live, you may also have legal rights to access your personal data, correct it, delete it, receive a portable copy, restrict or object to certain processing, and withdraw consent you previously gave. To exercise any of these, email us at the address in section 14. We answer every request, and we will not treat you differently for making one. If you are in the European Economic Area or the United Kingdom, you also have the right to complain to your local data-protection authority.

11. Security

All data is transmitted over HTTPS/TLS. Access to your workspace requires your signed-in account, requests to our backend are verified server-side rather than trusted from the app, and AI provider API keys exist only on our servers, never in the app or the browser bundle. No system is perfectly secure, but we keep the attack surface deliberately small: no analytics, advertising, or tracking SDKs, and no third-party code beyond the providers listed in section 6.

12. Children

Machina is not directed at children under 13, and we do not knowingly collect data from them. If you believe a child has created an account, contact us and we will delete it.

13. Changes to this policy

If this policy changes we will update the date at the top of this page, and point out material changes in the app rather than relying on you to re-read it. Sections 1 to 14 are the operative policy; the summary at the top is a plain-language guide to them and does not modify or limit anything below.

14. Contact

Questions or requests about your data: support@mymachina.app