Privacy Policy

Flowy is operated by Calcura Software, LLC. You can reach us at hello@tryflowy.app.

Account information. An email address, and an identifier from Apple or Google if you sign in with them. If you use Sign in with Apple with “Hide My Email”, we only ever receive Apple’s private relay address, not your real one.

Content you save. Whatever you share to Flowy: links, screenshots, photos, videos, PDFs, other files, and emails you forward to your Flowy address. We also store what we derive from it — titles, summaries, tags, categories, transcripts, and text recognised in images — along with any notes or tags you add yourself.

Optional personalization profile. Answers you choose to give about your work or learning, current goals, and preferences. We store them with your account so you can review and update them across your devices. Completing this profile does not enable automatic memory extraction from your conversations.

Preferences and delivery. Your daily digest settings, your choice about product news by email, your Flowy email alias, the language you created your account in (so our emails reach you in it), and, on mobile, a device push token so we can tell you when something you saved has finished processing.

Service usage and technical information. We record limited product events, such as which feature was used, the type and source of a saved item, and whether a workflow succeeded. When product analytics is configured, the analytics provider also receives a pseudonymous account identifier and browser, device, and connection information. Our defined product events do not include the content you save, titles, notes, files, URLs, emails, search questions, chat messages, or personalization-profile answers.

We do not use advertising trackers, we do not sell personal information, and we do not use your content to train AI models.

Flowy cannot do its job without sending your content to a few processors. Each one receives only what it needs:

• Anthropic — the text, images, and transcripts of what you save are sent to Claude to produce titles, summaries, tags, and to answer your questions in chat. When personalization is enabled, your declared profile is also sent as context to tailor those answers.
• OpenAI — extracted text is sent to generate embeddings (numeric representations), audio from saved videos is sent for transcription, and, when the direct OpenAI analysis route is enabled, the text of saved content is sent to produce titles, summaries, tags, categories, source-grounded structured results, and configured daily, weekly, or monthly digest reports. This route is disabled by default.
• Voyage AI — extracted text and chat questions are sent to create search embeddings and, when enabled, improve the ranking of search results.
• OpenRouter (OpenRouter, Inc., openrouter.ai) — added on September 23, 2026: used for accounts that accepted the disclosure dated September 23, 2026 or later, and for earlier accounts from September 30, 2026 (see “Changes to our AI providers” below). OpenRouter routes each request to the hosting provider it selects for an open-weight model (currently Qwen). We use it for two things: to produce titles, summaries, tags, and categories of the content you save, and, during evaluation, to generate a parallel result that we compare with Claude’s and never show you in its place. We send every request to OpenRouter with its data-collection setting set to “deny”, which excludes hosting providers that retain prompts or train on them.
• SocialVault — when you save a link from Instagram, TikTok, LinkedIn, Facebook, or Threads, the link is sent to fetch the post’s publicly available media and captions.
• Cloudflare — files you upload are stored in Cloudflare R2.
• Railway — hosts the application and the database.
• Expo — delivers push notifications to your device.
• Resend — sends the emails we send you. It receives your email address, your first name if you gave one, and the content of each email.
• PostHog — provides product analytics and error diagnostics. It receives the pseudonymous account identifier and the limited product events described above so we can understand aggregate use of Flowy. We disable automatic capture, which could otherwise collect the text of clicked elements, and do not record sessions inside the signed-in app. Session recording is limited to selected public pages and masks form inputs. Optional error tracking sends a redacted error type, technical code location, platform and app version. It excludes error messages, saved content, request bodies and credentials.

When you connect Flowy to ChatGPT or another compatible MCP client, you sign in to Flowy and approve the requested permissions. Read permission lets the connected client request matching saved items, original text, summaries, personal notes, transcripts, text recognised in images, and published digests from your library. Results can include private Flowy reader links. Returned content is sent to the client you authorised so it can answer your requests; that client’s own privacy terms apply to its handling of the content.

Separate write permission lets the client save an HTTP(S) link when you request it. Saving uses Flowy’s usual content processing and the AI providers described in this policy. Semantic searches send the search query to Voyage AI to create a search embedding. Reading stored items or published digests does not generate a new Flowy AI answer or digest.

Flowy stores the authorisation grant and associated client and permission information needed to operate the connection. Clients receive revocable MCP tokens, not your PocketBase credentials. Grants expire after 30 days; access tokens expire after 15 minutes. You can revoke a connection in Settings → Connections. Revocation stops further access through that grant; it does not remove content already shared with the client.

For help with MCP connections, email hi@tryflowy.app.

Account emails. These are part of the service: a welcome email when you create your account, password resets, important notices about your account or these policies, and the digests you choose to schedule. They stop when you delete your account, and you can turn digests off at any time.

Product news, only if you ask. With your consent we occasionally email you about new Flowy features, launches, and tips for using Flowy. You give it by ticking the optional, unchecked box when you create your account, or later under Settings → Digests → Marketing emails; creating an account alone does not subscribe you. We record when you gave or withdrew it. We only send news about Flowy, never offers from other companies, and we never share your address for anyone else’s marketing.

If you joined during the beta, before October 4, 2026, we included you in product news; the first of these emails says so.

You can stop product news at any time, in one tap, under Settings → Digests → Marketing emails, or from the link in any of those emails. Unsubscribing takes effect for every later send and does not affect account emails.

We use service usage and technical information to operate, secure, measure, and improve Flowy. For example, we look for aggregate patterns to understand which features are used, where a workflow fails, and which improvements to prioritize. We use those insights for our internal product and business planning; we do not sell personal information or use this information for advertising based on a person’s activity in Flowy.

We may create and use aggregate or de-identified statistics from that information only when they are not reasonably capable of identifying you. We do not try to re-identify a person from those statistics or combine them with other data to do so. A pseudonymous account identifier is not anonymous data, and we continue to treat it as personal information. Except as described below, we do not use your saved content, notes, files, emails, search questions, chat messages, or personalization profile to create these statistics.

To decide how to improve Flowy for each kind of content, Flowy staff may review the posts, reels and videos you save from public social and video platforms such as Instagram, TikTok, YouTube or Reddit — never profile or account pages — including the page address without its query parameters only when it contains no account name, its title with email addresses, phone numbers and @handles removed, its platform, the week it was saved, and the category Flowy assigned — under a pseudonymous label that is never shown next to your name or email. For any other website we only count the domain. This review never includes files, screenshots, receipts, audio, recordings, emails, notes, summaries, or chat messages, and we never use it for advertising.

Before we create an account or send your content to Anthropic, OpenAI, Voyage AI, or OpenRouter, we ask for your explicit permission. By checking the AI processing box during sign-up, you authorize those disclosures solely to provide Flowy’s summaries, transcription, search, and chat features, and to compare the quality of the models that produce them.

You may decline by not creating an account.

Which AI providers Flowy uses is Flowy’s decision. We may add, replace, or stop using an AI provider at our discretion, for example to improve quality, reduce cost, or keep the service running, without asking you for a new permission. Every provider that processes your content is always listed on this page, receives only what it needs for the purposes above, and may not train AI models on your content.

Before a newly added provider starts processing your content, we will tell you in the app or by email at least 7 days in advance and update this page. If you do not agree, you can delete your account at any time before the change takes effect, and we will destroy your data as described below. Continuing to use Flowy after the change takes effect means you accept it.

Accounts that accepted the disclosure before September 23, 2026 were told on that date that OpenRouter was being added; their content may be processed by OpenRouter from September 30, 2026.

When you save a public web page, Flowy keeps the text it extracted — the title, summary, tags, and article text — in a shared cache keyed by the page’s address. If someone else later saves the same public page, they get that text instead of us fetching and analysing it again. This makes Flowy faster and cheaper.

This cache holds only text derived from publicly reachable pages. It never contains your files, your notes, your tags, your personalization profile, or anything identifying you, and files you upload are never shared with another account.

We keep your content until you delete it. Deleting an item removes it from your library. Deleting your account destroys the account and everything attached to it.

You can pause personalization while keeping your answers, or clear your profile from Settings. Clearing removes the profile answers and turns personalization off. Existing chat messages, including any that mention those details, remain in your chat history.

Model evaluation records. While we are evaluating models, the text a model returns for your content — for example a summary or a digest draft — may be kept together with the request’s metadata (which model ran, how long it took, what it cost, and whether its output was accepted) so we can compare quality between models. Only Flowy staff can see these records. We remove the returned text within 30 days, and sooner when you delete the item or digest it came from or when you delete your account.

You can delete your account from Settings, in the web app and in the iOS app. Deletion is immediate and cannot be undone.

When you delete your account, we destroy:

• your account record, email address, and sign-in identifiers;
• every item you saved, with its summaries, tags, notes, and transcripts;
• every file you uploaded, from our file storage;
• your search embeddings, digests, personalization profile, and push token;
• any model evaluation records linked to your account.

If you signed in with Apple, we also tell Apple to revoke Flowy’s access to your Apple Account.

Text already contributed to the shared cache described above stays, because it is derived from public pages and is not linked to you after your account is gone.

You can access and correct your information in the app, and delete it at any time. If you want a copy of your data or have a request we can’t action in the app, email us and we will respond.

Flowy is not directed to children under 13, and we do not knowingly collect their data.

If we change this policy in a way that materially affects you, we will update the date at the top and notify you in the app.