Privacy Policy
Last updated 17 October 2026
1. What this covers
This policy explains what Frontpiece collects when you use the service, what we do with it, who else processes it on our behalf, and how to get it deleted. It applies to this website and to your account. In this policy "we", "us" and "our" mean Frontpiece LLC, a Texas limited liability company (filing number 1616516830003), which is responsible for the personal data described here and is its controller where that term applies.
It also covers businesses we email about Frontpiece. If we have written to you, we hold the business email address your company publishes, your name if it is published with that address, your company and brand name, the public listing we looked at, the sample kit we made from it, and, for the link to that kit, whether and when it was opened, which parts were viewed and for how long, and what was downloaded. We use these to send that email and at most two follow-ups, and to check the quality of the kits we make. If you reply "stop" or write to legal@frontpiece.ai, we add your address to a do-not-contact list the same day and keep that one entry so that we never write again; ask, and we will delete the rest, including the sample kit.
We are a small team. We collect what the product needs to run and not more, and there is no advertising business here to sell anything into.
2. What we collect
We collect these things, and only these things:
- Account details. Your email address, and either a password, which is stored hashed by our authentication provider and is never visible to us, or your Google account, if you choose Continue with Google. Google then tells us your email address, whether Google has verified it, your name, your profile picture and Google's id for your account, and our authentication provider keeps these with your account. We use the name and picture only to show which account is which on a device you have signed in on: when you switch accounts, and on the sign-up page for accounts used there in the last 30 days. Google never gives us your Google password, and we ask for no access to your email, files, contacts or anything else in your Google account.
- Content you upload. Product photos, brand logos, and any reference images you add to an edit.
- Content you enter. Product names, brand names, descriptions, features, dimensions, categories, target buyers, brand voice notes, the marketplaces you are listing on, and the instructions and prompts you write for a generation.
- Brand facts. Facts we find about your brand on public websites (the year it was founded and where its products are made), the page each one came from, and your answers when we ask you to confirm them. We use a fact in your images and copy only after you confirm it.
- Content you import. When you import a product from a marketplace listing you supply, we fetch that public page and its images.
- Content we generate for you. The images, listing copy and A+ modules produced from the above, and their history: each version of an image or module, the instruction you wrote for an edit and any reason you picked for it, which editing tools you used when you saved a change without typing an instruction, which version you kept, and each change you make to your listing copy together with the version before it.
- Payment details, handled by Stripe. Card numbers never reach our servers. We store a Stripe customer reference, your subscription state, your plan, and the billing period.
- A record of your acceptance of these documents. When you tick the box to accept our terms of service and this policy, we record the date and time, the IP address you accepted from, your browser's user agent string, the version and a content fingerprint of the exact text you were shown, and the wording of the notice itself. This record is what allows us to show, later, that the agreement was actually formed and on what terms. Section 7 explains how long we keep it and section 8 explains why it is the one thing that outlives your account.
- Technical and error data. When something breaks we receive an error report that can include the page, the request and a stack trace.
- Analytics. PostHog receives content-free page and feature events, finite choices, counts and opaque account or object IDs so we can understand which features get used and where flows fail. This includes numeric page-performance measurements and safe error classifications. Checkout analytics can include the selected plan or pack and its listed price, but not card details, Stripe identifiers or payment ledger records. Account profiles use coarse plan, usage, marketplace and category buckets; unavailable information stays unknown. We do not send it your photos, prompts, listing content, names, email addresses, input values, element text or attributes. Session replay is off everywhere, and PostHog is configured to discard client IP addresses. In the European Union, European Economic Area, Switzerland and unknown locations, optional product analytics stays off until you accept on the cookie notice. In the United States, the United Kingdom and other regions it starts on without a notice, and you can switch it off at any time, free, with Cookie preferences in the site footer, at the foot of a shared kit page, or in the account menu. Declining or switching analytics off in Cookie preferences stops collection and clears analytics storage on this device. Global Privacy Control keeps analytics off, including when you accepted previously. Our host provides the connection country to choose the regional default; the region cookie holds only that default until you close the browser. Our host separately counts pageviews without a cookie.
- Records of what the service did for your account. For example, an image request that was refused or capped and why, where an added image came from, the result of removing a background from a logo, which resolution preset you set, which images and A+ modules you downloaded and from where, how many times a share link was opened, when it was first opened, and how many times images were downloaded from it. For a share link we also note whether it was you who opened it while signed in. In these records we keep nothing about anyone else who opens or downloads from it: no IP address, no browser details, no name; a visitor's own choice about analytics, described above, still applies to their visit. We keep these on our servers whatever you choose for analytics, we use them to run and improve the service, and we never sell them or send them to an analytics vendor.
3. Who processes it
We use the service providers below to run the product. Each one receives only what its job requires, and each processes it on our instructions.
- Supabase: accounts and authentication, the database holding your products and runs, and the private storage holding your uploaded photos and logos and the images we generate for you.
- OpenAI: image generation. Your uploaded product photos and your instructions are sent to its image API.
- Google: Gemini analyzes uploaded product photos before generation and powers in-product suggestions, so those photos and prompts are sent to it. Until September 2026, Google Drive also stored the generated images, and it may still hold images made before then.
- Anthropic: parses the marketplace listing pages you import, to fill in product fields. It also looks up public facts about your brand on the web, such as the year it was founded and where its products are made, using your brand name and listing text, so we can ask you to confirm them. It also sorts the edit instructions you write and the changes you make to your listing copy into categories, such as a wrong size or a wrong background, so we can see which parts of the service need fixing. We remove email addresses, links and phone numbers from that text first, and we send it no images.
- n8n: orchestrates the generation workflow, and passes your product details and photo links between the steps.
- Trigger.dev: runs background jobs such as adding an image, editing one, and upscaling.
- Vercel: hosts and serves the application.
- Stripe: payments, subscriptions and the billing portal.
- Resend: delivers account email such as password reset links, and receives your email address in order to send it.
- Sentry: error monitoring, which receives diagnostic data when something fails.
- PostHog Inc.: content-free product analytics on servers in the United States, only while analytics is allowed as section 2 describes. Session replay is disabled. When analytics is off, PostHog receives no optional analytics events from your browser or our servers. The first-party service event rows described in section 2, Stripe identifiers and full payment ledger records stay in our service records and never go to PostHog. The bounded checkout choices and listed prices described in section 2 can go to PostHog only while analytics is allowed.
- Replicate: image upscaling and background removal. It receives the image or photo needed for the operation and returns the processed image.
- Cal.com: the Book a call scheduler. Pages that offer it load Cal.com's booking script, which receives your browser's standard request details, and if you book a call Cal.com receives the name, email address and time you enter.
- Marketplace data providers: OpenWebNinja, Apify, Keepa and Etsy's API. When you import a product or use the free listing grader, they receive the listing link or product identifier you gave us and return that public listing. When we research the listings that sell best in your category, they receive search words describing your product and return public listings from the marketplace. They receive no photos, account details or payment information.
4. Artificial intelligence providers and training
Your content is sent to the artificial intelligence providers named above so the service can do its job. It is processed under those providers' business and API terms rather than their consumer terms. Under the OpenAI and Anthropic API terms, content sent through the API is not used to train their models. The Gemini API is used on a paid tier, and under Google's terms for that tier your content is not used to improve or train Google's models.
We do not train any model on your content ourselves, and we do not license your content to anyone to train theirs.
Using your edits to improve our own instructions and rules, as section 6 describes, does not train any model.
Generation is not deterministic and the providers may change their models at any time, so the same input can produce different output on different days.
5. Sharing, and links that do not need a password
We do not sell your personal information and we do not share it with anyone for their own marketing.
Two parts of the product make content reachable by a link rather than by a login, and both are worth understanding before you use them:
- Share links. When you share a kit, we make an unguessable link that lets whoever holds it view and download that kit's images, listing copy and any A+ modules. It needs no account and no password, and it does not expire on its own. You can turn it off at any time with Turn off link in the Share panel, which also closes any older link to the same kit; anyone who opens it afterwards sees that it is no longer active. Treat a link like a password, and only send it to people you mean to give access to.
- Image links. Generated images are served from storage links that work for anyone holding the URL. That is what lets a share link, an export and a shared preview display them without a login.
6. What we use it for
To run the service: to generate images and copy for you, to show you your work, to let you edit and export it, and to share it with the reviewers you choose.
To bill you, to prevent abuse, to diagnose failures, and to answer you when you contact us.
To improve the service. We study the records described in section 2, the edit instructions you write, the reasons you pick and the changes you make to your listing copy, to find where our images and copy go wrong and to fix our instructions, rules and product import for everyone. We look at them as counts across everyone; any short example we keep comes only from our own team's use of the product, never another seller's words. We never show them to other customers, and we never use them to advertise to you.
To comply with the law, and to enforce our terms. This is also why we keep the acceptance record described in section 2: it is the evidence that our agreement with you exists, and it is kept on that basis rather than to profile you.
We do not use your content to advertise to you, we run no advertising trackers, and we sell nothing about you to anybody. We do run product analytics, and section 2 says what that means: our host counts page views without a cookie, and PostHog receives the content-free events described in section 2 while analytics is on for you. We use it to find what breaks and what nobody can work out how to use, and for nothing else.
7. How long we keep it
Your account content, meaning your products, photos, runs, generated images and listing copy, is kept for as long as your account is open, because it is the working history the product is built around.
When you delete your account we delete that content, other than anything we must keep to meet a legal, tax or accounting obligation. Billing records are kept for as long as the law requires, usually several years.
The acceptance record described in section 2 is kept for six years after it is superseded by a later acceptance or after your account closes, whichever is later. That is a little longer than the period in which a claim about the agreement can still be brought where these terms place disputes. It is stored so that it cannot be edited or deleted by the application, which is the property that makes it worth anything as evidence.
Error reports are kept on a short rolling window by our monitoring provider, and are not linked to your content.
The records described in section 2 of what the service did for your account are deleted after 24 months, or when your account closes, whichever comes first.
Content held by the providers in section 3 is retained under their own terms. Our PostHog Free plan retains product analytics events for one year. Existing recordings have a 30-day retention setting; new session recordings are disabled. The twelve-month analytics cookie duration is a device-storage duration, not an event-retention period. You can contact us to request deletion of optional analytics linked to your account.
9. Security
Uploads live in a private storage bucket, database access is scoped so an account can only reach its own rows, and traffic runs over TLS. Payment card data never touches our servers.
No service is perfectly secure, and section 5 describes two places where the product deliberately trades a login for a link. If you become aware of a vulnerability, tell us at the address below before you tell anyone else.
10. International transfers
We are based in the United States and the providers in section 3 operate there and elsewhere. Using the service means your content is processed in the United States and in other countries where those providers run, which may have different data protection rules than your own.
11. Children
The service is for businesses and is not directed at children. You must be at least 18 to use it, and we do not knowingly collect anything from a child. If you believe a child has given us information, tell us and we will delete it.
12. Changes to this policy
We may update this policy. When we do we will change the date at the top of this page. Where a change is material we will ask you to accept it before you continue to use the service, in the same way as a change to the terms, and we will keep a record of that acceptance.
13. How to reach us
Privacy questions, data requests and deletion requests all go to the same place: legal@frontpiece.ai. The terms of service set out the agreement this policy sits under.