Privacy

Last updated 26 October 2026. Propozit is in early access; this page covers what we collect today.

Early-access requests

Before anyone could create an account, people could ask for early access with a form. For those requests we keep the name, work email, company and business type they gave. We use them only to contact them about Propozit and to understand who it is for. We don't sell them, and we only share them with the processors listed below. Ask us to delete them at any time and we will.

Accounts

When you create an account we store your name, email, position and the organizations you belong to, plus what you enter in settings (company details, brand and logo). Admins can export your clients and their contacts as a CSV file. Sign-in uses one-time codes sent by email; we don't store passwords. To limit sign-ups from one place, we keep the network address each sign-up comes from (for IPv6, only its first half) for up to two hours. If you add a signature, it is stored privately: only you can open the file, and it is shown only on the agreement slide of proposals where you are the contact and in their signed PDFs.

Proposals your clients open

When a client opens a proposal, we count the open. To open a private proposal, or to sign, they enter a code we email them; that unlock is stored with their email address, network address and browser type (user agent), and the first time it happens the sender's contact person gets an email saying which address opened it. Unlocking with an email code always identifies them to the sender, whatever they choose below. A small bar on the proposal asks whether the sender may also see which slides they read and for how long. Only after they choose OK do we store how long each slide was on screen and link their later opens to their email address, using a first-party cookie that tells their visits apart and lasts one year. Before they answer, or after No thanks, the open itself is counted without that cookie, their email address or any other identifier. Their answer is kept for that proposal in a cookie for one year. One more first-party cookie keeps them signed in to that proposal for 30 days once they unlock it with an email code (on a public proposal, only if they confirm their email to sign or to ask for a change). None of these cookies is used for anything else, and no third party sees them. The organization that sent the proposal sees this in its activity panel. When they sign, we store their name, role, signature, email address, network address, browser type (user agent) and the time, with the signed PDF. When they ask for a change, we store their email address, the topic they chose, their message (up to 2,000 characters) and the time, for as long as the proposal exists, and email it to the proposal's contact person (or, without one, to up to five owners and admins of the sending organization).

Bank details and payments

This section is about your clients paying you. Paying for Propozit itself is under Paying for Propozit. Admins can add bank details under Settings, Payments: an account holder, an IBAN, a BIC, a PromptPay ID and a pay link. They are business data your team enters, and a sole trader's IBAN may be personal data. After a client signs, they are shown to the person who signed: on the proposal page, in the signed email, in the signed PDF and in payment requests. The signed PDF, which includes your bank details when you set them, can be downloaded by the person who signed and by the people the proposal was shared with. Strangers with a public link do not see them. When a client says "I've paid", we store that claim, and when a team member marks a payment as paid or unpaid we store the paid date and who did it. Owners and admins get an email when bank details change; it shows only the last four characters of the IBAN. Propozit does not handle the money and does not check payments. No new processor is involved.

Webhooks you set up

Owners and admins can add webhook URLs under Settings, API and webhooks. When something happens to a proposal, we post an event to those URLs: the event type, the proposal's title and link, the client company, and for some events the name and email of the person who signed, declined or asked for a change, or the email addresses it was sent to, with the reason or message they wrote. Payment events carry the amount, the currency and the payment reference. The organization chooses these URLs and is responsible for them and for what happens to the data there. They may be outside the EU. Each event is signed, and only sent over https. We keep each event's data for 30 days, so it can be tried again or resent, then delete it. The delivery log for each URL holds the time, the event type, the result and the error, never the data itself, and goes with the event after 30 days. API keys are stored only as a fingerprint; we keep each key's name, who made it and when it was last used, and count an organization's requests per day for 7 days.

Where data lives

We use these processors: Supabase stores the database and handles sign-in (EU), Amazon Web Services hosts the application (EU region), Resend sends our emails and Stripe handles payment for paid plans (see below). For now, our AI features use Claude by Anthropic, through Anthropic's API, which processes the data in the United States; the sections below say what is sent. We use strictly necessary cookies to keep you signed in, to remember your last organization, and on client proposals as described above; there are no advertising or third-party cookies.

Paying for Propozit

When the owner of an organization chooses a paid plan, payment is handled by Stripe: the Stripe company named in our agreement with Stripe. When the owner starts a checkout, Stripe receives the organization's name and the owner's email, even if the checkout is not finished. To pay, it also receives the billing address, the VAT or tax ID if you give one, the country and the card. You type the card on Stripe's own page: Propozit never sees or stores the card number. Stripe may transfer data to the United States under its own safeguards, and it keeps its own records as an independent controller for its own legal duties, such as preventing fraud and meeting financial rules, as its privacy policy describes. We keep the Stripe customer and subscription IDs, the plan, the billing period, the subscription's status and its currency with the organization, so we know which plan applies. Invoices, the card and the billing address stay with Stripe; the owner reaches them through Manage payment and invoices. If a payment fails, the owners get an email from us. Nothing goes to Stripe until an owner starts a checkout.

Setup from your website

When you set up from your website, Propozit reads the public pages of the address you enter, from our servers in the EU. It keeps what it found, with copies of the images it may suggest, while you review it. When you finish setup, it deletes what it found and removes those copies right away; if removing a copy fails, it goes with the 30-day clean-up below. Of what it found, it keeps only a record of which suggestions you saved. Reading your website again removes the copies from earlier readings. Any reading that is more than 30 days old is deleted with its copies; that clean-up runs each time anyone starts reading a website, so it can come later when nobody does. It does not log in, follow other websites or send anyone an invitation. Only those public pages, and images from them, are sent on; your own account data is not. Two more processors help with them:

• Anthropic (Claude Haiku, through Anthropic's API) reads the text of those pages to find details such as prices and your company address. Anthropic's API processes this data in the United States. Anthropic does not use it to train its models. We plan to move AI processing to an EU provider.
• Subquadratic (SubQ Snap) receives short snippets of those pages and up to four images that may be your logo, to rank the suggestions. Subquadratic does not publish a data region, so we cannot say where it processes data and it may be outside the EU.

The proposal AI

When you describe a proposal and let the AI draft it, your description goes to Anthropic (Claude Haiku, through Anthropic's API), with the proposal data it needs: your organization's name and type, your products and prices, deals and bonuses, add-on people's roles and rates, your cases (titles, clients, tags, taglines, summaries, numbers and results), client names and the template texts. Anthropic's API processes this data in the United States. Anthropic does not use it to train its models. We plan to move AI processing to an EU provider. Nothing goes to Subquadratic (SubQ) or Jev for this. Nothing is sent until you choose to draft a proposal or ask for changes.

When you ask the AI for changes, your message, up to the last six messages of that conversation, the proposal's text, its price lines (names, options and quantities), your products and prices, and your cases go to Anthropic in the same way. Nothing is sent until you select Ask. We don't store your messages or the AI's answers; we count each request for the daily limits, and our logs hold counts and timings only.

We keep your description and the AI's plan with the draft for 30 days, then delete them, together with the things it could not find and any new client name it suggested. After that we keep only what it chose from your account (also in your proposal) and how much it was used, to count the daily limits. Our logs hold counts and timings, never what you wrote.

Contact

Questions or deletion requests: hello@propozit.com.