Privacy

Flowpage is built and run by Lemonstone AI in Dubai. Flowpage is the working name of the product while it is in beta.

Two different people are covered here, and what we hold about them is not the same. A broker has an account and gives us material to work with. A buyer has no account and never signs into anything. They were sent a link and opened it.

If you have an account

What we hold:

• Your name, your email address and the handle your pages live under.
• Everything you upload, meaning brochures, fact sheets, price lists, images and links, and everything read out of them.
• The pages you build, and a saved snapshot of every version you publish.
• Your brands, your designs and your layouts.
• Your credit history: what each page cost and when, so the balance on your Settings screen can be a real number rather than a guess.
• The figures about the pages you published. See the next section.

We use it to build your pages, serve them and keep your account working. We do not sell it, we do not hand it to advertisers, and we do not use it to sell anything to your clients.

Who else touches it

Three companies, each for one job:

• Anthropic, the AI that reads your documents and writes the page from them. Your documents and their text are sent there. They are not used to sell you anything and they are not shared with other accounts.
• Supabase, the database and the file storage. Everything above lives there.
• Vercel, the servers this app and your published pages run on.

There is no analytics service, no advertising pixel and no third-party script anywhere in this product, not in the app and not on a page a buyer opens. A published page loads nothing from anyone but us, fonts included.

If you opened a page somebody sent you

You have no account here and nothing has been signed up for. The page records that it was opened, so the person who sent it can see whether it was read. There is no notice on the page and you are not asked anything. This section is where that is written down instead.

One cookie, called _fpv

Opening a page sets a single cookie called _fpv. It holds 128 random bits and nothing else: it is not built from your device, your browser, your address or anything about you, so there is nothing in it to work backwards from. It is marked HttpOnly, which means the page's own scripts cannot read it either.

It is one cookie for this whole website, not one per broker. So if you open pages from two different brokers here, both opens carry the same random id, and each of those brokers can see that the same anonymous visitor came back to their own page. Neither of them sees the other's pages, and neither of them learns who you are. This is the ordinary way a first-party cookie works, and it is a deliberate choice rather than an accident.

Clearing your cookies clears it. The next page you open mints a new one, and you look like a new visitor.

What is recorded

• That the page was opened, and when.
• How long it stayed open, how far down it was read, and which sections were reached.
• Which link was used, if the broker made a named one. A name is attached only in that case, because the broker typed it themselves when they made a link for one person. An ordinary link carries a count and nothing else.
• Whether the open was a robot rather than a person, such as a link preview, a crawler. A robot is recorded so the broker can tell it apart from a reader, and it is never given a cookie.

What is deliberately not kept

• Your browser's user-agent string is not stored. It is read once to decide whether the open was a robot, that yes-or-no is kept with a short reason, and the string itself, which names your device, your operating system version and your browser build, is thrown away.
• A referring address is kept only as its origin. If you arrived from WhatsApp we keep https://wa.me and drop everything after it, which is where a conversation id or a search term would otherwise sit.
• Your IP address is not stored. It is turned into a one-way scrambled value with a secret key, which is used to spot duplicate traffic and to stop one machine flooding us. The address itself is not written down. The copy that counts requests, so that no one machine can flood a page, stops counting a minute after it is made; the row is then cleared away the next time a request arrives from the same address, or by a periodic sweep.
• No name, no email, no phone number, no location, no advertising identifier.
• Nothing at all is read out of the page, and nothing is stored on your device beyond the one cookie.

How long any of it is kept

For as long as the account it belongs to exists. While this is a beta, nothing is deleted on a schedule. The figures about a page are kept for as long as the page is, and the page for as long as the account is. When the beta ends, this is one of the things that will be given a real answer. The one exception is the scrambled address described above: the copy used to count requests carries its own expiry, about a minute, after which it counts towards nothing. The row itself goes on the next request from that same address, or in a periodic sweep.

Deleting your account

Settings → Delete account. It destroys everything and there is no undo: your pages and every published version of them, the pictures and documents you uploaded and their files, your brands, designs and layouts, your links, your credit history, and the figures recorded about your pages. Published links stop working.

Nothing else destroys anything. Not using the product, letting the account go quiet, or the beta ending removes nothing on its own. Deleting the account is the only thing that does.

One thing we cannot delete yet, and we would rather say so than leave it out. When a document is read, a copy of it is uploaded to the AI provider's file service. We have no way to delete those copies today. Everything on our own side goes; that copy does not, and building the way to remove it is on the list.

Asking us about your data

If you have an account, most of it is on your own screens and the delete button is yours. For anything else, such as a copy of what we hold, a correction, or a question about a page somebody sent you, the quickest way to reach us is the contact page at lemonstone.ai.

Two things are still to be filled in before this product is offered outside the invited beta: the full legal name and registered address of the company behind it, and the formal basis on which the data of somebody in Europe would be handled. They are named here as missing rather than answered vaguely.