How Veram handles personal data.

Effective 26 October 2026. The plain-language version is on Your data; if the two ever disagree, this policy governs and we will fix the other.

1. Who we are

Veram is a service of Promvie, Inc., a Delaware corporation, 1111B S Governors Ave STE 28868, Dover, DE 19904, United States (“Veram”, “we”, “us”). For the purposes of the GDPR and similar laws, Promvie, Inc. is the controller of the personal data described here. Contact us about anything in this policy at hello@veram.io.

2. What we collect, and why

We collect only what the service needs. For each category: what it is, where it comes from, what we use it for and, for people in the EU and UK, our legal basis.

• Portrait text. The text your AI assistant writes about you, which you paste into Veram (or which your assistant submits through our connector). We read it once to build your profile and then delete it; connector submissions are held for at most 24 hours, only until you open the link. We never connect to your AI account or see your conversation history. Basis: contract.
• Your profile. A headline, a narrative, a set of claims about you and, for each claim, your AI’s short note on how it knows it, with approximate dates (portraits made before 21 October 2026 keep short quotes of your own words instead); counts of what our filters set aside. Basis: contract.
• Your checks and answers. Which claims you mark right, partly right or wrong; your first name, gender, date of birth, country and state or region; who you’re interested in and an age range; your answers about how you live and relate (a good week, how you handle disagreements, kids, relationship shape, smoking, where you’ll be, the next two years); a line on what people get wrong about you; optional dealbreakers; if your portrait is short, optional lines on what you did last Saturday, a plan you have, and the last time you went out of your way for someone. If you tell us after a few passes what your next introductions should have more of, we keep that choice. Basis: contract.
• Where you live. Your zip code, the place and state it belongs to, the centre point of that zip code (from GeoNames, geonames.org, used under CC BY 4.0), and how far away you want introductions to be. If you tap “Use my location”, your device’s location is sent once to find the nearest zip code and is not stored. We never keep your own coordinates. Basis: contract.
• The basics (optional). Your height, religion, political views and whether you drink, only if you choose to add them. They are shown on your introduction card and are not used to choose or score who you meet. You can clear any of them at any time. Basis: explicit consent, given by adding them.
• Photos. Up to six photos you upload. We re-encode them, remove embedded location and camera data, and check each one automatically against our community guidelines before any other member can see it; a photo that breaks them is not stored. Basis: contract; legitimate interests (keeping members safe).
• Selfie for verification (optional). One camera frame taken while you perform a random gesture. It is compared with your first photo automatically, by an AI model, and the selfie is deleted as soon as the check is decided (if the model can’t be reached, a member of our team compares them instead). If you later remove the photo it was compared with, the verification is removed too. We do not create face templates, face geometry or any other biometric identifier from your photos or selfie. Basis: consent.
• Deep profile. Scores on a set of scales (what you value, where your life is going, temperament, how you relate, daily rhythms) and a conflict style, each with a confidence and, for portraits made before 21 October 2026, where safe, one quote of yours. Built by our models from your profile and answers; optionally refined by your own AI when you paste its scoring (that text is read once and discarded); and by your own answers. Basis: contract.
• Steadiness and closeness (optional, off by default). Three scores estimating emotional steadiness and two attachment tendencies. Because they are inferences that can relate to mental well-being, we treat them as sensitive data and as consumer health data (see section 12). Computed only after you switch them on. Basis: explicit consent.
• Matching data. Text embeddings of your profile; compatibility scores; for people who already pass each other’s filters, a model’s estimate of how well two profiles fit, with a one-line reason (never shown to either person); introductions and the written reports about them. Basis: contract.
• Your activity with other members. Whether you chose to meet or pass (and the reason you picked, if any); messages you send in a mutual thread, which are checked automatically for signs of fraud, sexual exploitation, minors, threats and harassment (a flagged message, and the conversation around it, is read by our team); photos you send in a thread, which are stripped of their metadata (including location) and checked automatically for nudity, anyone who may be under 18, children and violence before the other person can see them (a refused photo is not delivered and is handled like a refused profile photo); voice messages you send in a thread, which are turned into text by an automated model only so their words can get the same check as a written message (the text is not kept; if one is flagged, its words go to our team with the case); video calls you make in a thread, of which we keep only who called, when, and how long it lasted (the call itself goes directly between the two phones and is never recorded); delivery and read receipts, and when you last had Veram open; whether you met; conversations you muted; people you unmatched or reported, and why. Basis: contract; for safety records, legitimate interests.
• Comparisons. If you compare portraits with someone you know, the invitation, the names you each give, and the report. Basis: contract.
• Payments. When you buy reveals, our payment processor Stripe collects your card or wallet details directly; we never see or store full card numbers. We receive the purchase amount, currency, a transaction reference and your email. We keep a ledger of the reveals you receive, buy, use and get back. Basis: contract; legal obligation for tax and accounting records.
• Account data. Your email address; if you sign in with Google, the email and account identifier Google shares with us (never your password or contacts); your founding-member number. Basis: contract.
• Notifications. If you allow push notifications, the push address your browser gives us; a record of the emails we sent you, so we never send the same news twice; your email preference. Basis: contract; consent for push.
• Technical data. IP address, device and browser information and request logs, kept briefly by our hosting provider for security and abuse prevention; a one-way hash of your IP to rate-limit connector submissions. Basis: legitimate interests (keeping the service secure).
• Analytics. Events about how Veram is used (for example, that a page was viewed or a step completed), never your portrait, answers, photos or messages. Without your consent this is cookieless and cannot follow you across visits; with your consent we set one analytics cookie. On our public pages (the home page, the New York page, How it works, the FAQ, the founding members page, our comparison with Hinge and Tinder, the Journal and the first steps of making a portrait) we also record how a visit goes: what was on the screen, scrolling and taps, so we can see where people get stuck. Anything typed or pasted is hidden in these recordings, and the step where you paste your AI’s answer is left out. Signed-in screens and private links are never recorded, and nothing is recorded in a browser once it has been signed in. Where we ask about cookies, recordings happen only if you accept them, and never with a Global Privacy Control signal. Basis: consent for the cookie and, where we ask, for recordings; legitimate interests for cookieless counts and, elsewhere, for recordings of public pages.
• Ad measurement. We advertise Veram on Instagram, Facebook, TikTok and Google (including YouTube). In the United States, the Meta and TikTok pixels and the Google tag on our site tell those companies when a visitor takes one of these steps: starts or makes a portrait, keeps it in a new account, finishes a profile, reveals someone, starts a checkout or buys reveals (with the price, for checkouts and purchases). The TikTok pixel can also record by itself the buttons and links you tap, with the words on them and where a link goes. With each step they receive the page address, your browser and device, your IP address and a cookie each of them sets. If you are signed in, they also receive your email address and Veram account ID in hashed form: turned into a one-way SHA-256 code in your browser before it is sent, which they use to match the step to an account you may have with them. This lets them count which of our ads worked and show our ads to people likely to join. We never send them your portrait, answers, photos, messages, name, gender, who you are looking for, or anything about your health, and signed-in screens send no page views. You can turn this off (see section 15). It never runs in the EU, the UK or Switzerland. Basis: legitimate interests.
• Enforcement records. Warnings, suspensions and bans on your account, and the reason for each. If your account is banned, we keep a one-way fingerprint (hash) of your email address, your Google account ID, fingerprints of the payment cards you used (from Stripe, never the card number) and fingerprints of your photos, only to stop a banned account from coming back. Basis: legitimate interests (keeping members safe).
• Support, reports and safety checks. What you send us by email or through our removal request form, anything a member reports about you, what the automatic photo and message checks flag, and our review of it. Basis: legitimate interests; legal obligation where relevant.

3. What we keep out

Before anything from a portrait is stored, automated filters remove content about other people in your life (partner, family, friends, colleagues) and content about your finances, health, religion, political views or sexual orientation, along with names of employers, companies or real people. The filters are not perfect; if something slips through, tell us and we will remove it. Your religion and political views are kept only if you add them yourself to the basics on your card. We do not ask for your race, ethnicity, immigration status or precise location, and we do not collect your contacts.

4. How matching uses your data

Matching is automated profiling. We filter people by the hard lines both of you set (age and gender preferences, region, kids, smoking), score the rest on your deep profile, answers and profile text, have a model assess the strongest candidates, and write a report for each introduction. This decides whom you are introduced to and in what order. It does not produce decisions with legal or similarly significant effects, and a person can always review how an introduction was made if you ask. You can see and correct the inputs (your claims, answers and deep profile) at any time.

5. What other members see

• Before anyone chooses to meet: a match percentage, a few shared themes, each shown with a line from each of your portraits or answers, one area of friction, whether you are verified, roughly how many miles apart you live (never your zip code or town), and any basics you chose to add. Never your name or photos. A claim you marked wrong is never used. A line in your own words (from your answers, or from a portrait made before 21 October 2026) appears only if it passes an automated check for anything about other people, anything identifying and anything sensitive; and you can hide any one claim’s words, or all of them, from the claims screen.
• After someone chooses to meet you: your first name, photos, headline and portrait narrative, and the full report about the two of you.
• If you both choose to meet: the messages and photos you exchange, and delivery and read receipts.
• In a comparison you start or join: the other person sees the shared report, including lines from your profile and answers.

Your deep profile scores, your steadiness and closeness scores, your answers about hard lines (other than their effect on who you are introduced to), pair assessments and anything you told us in a report are never shown to other members. Nothing about you is public or searchable.

6. Who we share data with

We use service providers who process personal data only on our instructions and under written agreements:

• Vercel Inc.: hosting, delivery and background jobs (United States).
• Supabase Inc.: database, file storage and sign-in (United States).
• Anthropic PBC: AI models that read portrait text, build profiles and deep profiles, assess pair fit, write reports, and check photos and messages for safety. Under our agreement, inputs and outputs are not used to train Anthropic’s models.
• OpenRouter, Inc. and TypeSafe: a first, automated safety check on messages, and on lines in your own words before one may appear in a pair report. OpenRouter passes each message or line to TypeSafe’s model, we require zero data retention on every request, and TypeSafe commits not to train on them. Anything it is unsure about is checked again by Anthropic’s model. OpenRouter also passes voice messages to a Google model that turns them into text for that check, with zero data retention required.
• Stripe, Inc.: payment processing. Stripe is an independent controller of the payment data it collects for fraud prevention and legal compliance, under its own privacy policy. Where checkout shows the sale is made through Stripe’s Link service, Stripe is the merchant of record for the payment and collects your name and billing address to calculate sales tax.
• Resend, Inc.: sending email.
• Cloudflare, Inc.: relaying a video call between two phones when a network won’t let them connect directly. The call is encrypted end to end between the two phones, so Cloudflare can’t see or hear it, and nothing is recorded.
• PostHog, Inc.: product analytics and recordings of our public pages (United States).
• Linear Orbit, Inc.: the tool our team uses to track and review safety reports and flags (United States).
• Google LLC: sign-in, only if you choose it.
• Meta Platforms, Inc., TikTok Inc. and Google LLC, for ad measurement in the United States, as described in section 2. For that data they act as independent businesses under their own terms and privacy policies, not as our service providers.
• Your browser’s push service (for example Apple, Google or Mozilla): delivering push notifications you allowed; notification text is kept short and never includes your portrait.

We may also disclose personal data (a) to comply with law, a court order or a valid legal request; (b) to protect the safety of any person, investigate reports, or prevent fraud and abuse; (c) to enforce our terms; or (d) as part of a merger, acquisition or sale of assets, in which case the recipient will be bound by this policy for the data transferred and we will tell you.

We never sell your portrait, answers, photos or messages, and we never sell any personal data for money. The ad measurement in section 2 may count as “sharing” for cross-context behavioural advertising, or as “targeted advertising”, under some US state laws; you can opt out of it in section 15. We show no ads inside Veram.

7. Where your data is, and transfers

Our systems are in the United States. If you use Veram from outside the United States, your data is transferred there. For people in the European Economic Area, the United Kingdom and Switzerland, we rely on the EU–U.S. Data Privacy Framework (and its UK and Swiss extensions) where our provider is certified, and otherwise on the European Commission’s Standard Contractual Clauses and the UK Addendum, together with supplementary measures such as encryption.

8. How long we keep it

• Portrait text and text your AI writes to deepen your profile: deleted once processed; connector copies within 24 hours.
• Selfies for verification: deleted when the check is decided (usually within seconds; within a day if a person has to look).
• Steadiness and closeness scores: until you switch them off or delete your account; deleted immediately either way.
• Profile, answers, photos, deep profile, introductions, reports and messages: while your account exists.
• Results of the automatic safety checks: while your account exists.
• Pair assessments: while both accounts exist; recalculated when either profile changes.
• Records of reports and blocks, and enforcement records: up to 3 years after the account involved is deleted, to protect other members and to deal with legal claims.
• Ban-list fingerprints of a banned account: up to 5 years after the ban.
• An image that appears to show child sexual abuse: kept in locked storage, never shown to anyone, for as long as the law requires after we report it to NCMEC (at least one year).
• Payment and reveal records: 7 years, as required for tax and accounting.
• Security logs: up to 30 days. Analytics events: 12 months. Recordings of public pages: 30 days.
• Backups: deleted data may remain in encrypted backups for up to 30 days before it is overwritten.

9. Security

Data is encrypted in transit and at rest. Photos and selfies are kept in private storage and served only through links that expire within an hour. Database access is limited to your own records by row-level security; processing that spans accounts runs on our servers with keys that never reach a browser. No system is perfectly secure; if a breach affects you, we will notify you and the authorities as the law requires.

10. Your choices and rights

Wherever you live, you can:

• see and correct most of your data yourself on your portrait page;
• download a copy of everything we hold, or delete your account and everything in it, in one tap on Your data when signed in;
• switch emails, push notifications, analytics cookies and recordings, ad measurement and the steadiness and closeness scores on or off;
• ask us to access, correct, delete or port your data, or to explain how an introduction was made, by writing to hello@veram.io.

We answer requests within the time the law of your place of residence requires (for example 30 days in the EU and UK, 45 days in US states, extendable once where permitted). We may need to confirm your identity, normally by asking you to write from the email on your account. You may use an authorised agent where the law allows; we may ask the agent for your signed permission. If we decline a request, you can appeal by replying to our decision with “Appeal” in the subject line; we will respond within the time the law requires and tell you how to contact your state attorney general or data protection authority if you disagree. We will never treat you differently for exercising a right.

11. US state privacy rights

Residents of California and of other states with comprehensive privacy laws (including Colorado, Connecticut, Delaware, Oregon, Texas, Utah and Virginia) have the rights described above: to know and access, to correct, to delete, to portability, and to opt out of sale, targeted advertising and profiling in furtherance of decisions with legal or similarly significant effects. We do not sell personal data for money. Our ad measurement (section 2) may count as sharing or targeted advertising under these laws: you can opt out of it in section 15, and we treat a Global Privacy Control signal as a valid opt-out. Our profiling does not produce decisions with legal or similarly significant effects.

Notice at collection (California). In the last 12 months we collected: identifiers (email, account ID, IP address); personal characteristics you give us (gender, date of birth); commercial information (reveal purchases); internet activity (usage events); approximate geolocation (your zip code and its state); photographs; profile inferences (your portrait, deep profile and compatibility scores); and, only with your consent, sensitive personal information (account login credentials handled by our sign-in provider; your religion, if you add it to your card; and the steadiness and closeness scores, which may reveal health-related inferences). We collect these for the purposes in section 2, disclose them to the service providers in section 6, share identifiers (including your email address and account ID, in hashed form), internet activity and commercial information (the price of a purchase) with Meta, TikTok and Google for ad measurement, and keep them for the periods in section 8. Sensitive personal information is never shared with them. We use sensitive personal information only to provide the service you asked for, as permitted by the CPRA regulations, so no “limit the use” request is needed; you can still clear your religion or switch the steadiness and closeness scores off at any time.

California’s “Shine the Light” law: we do not share personal information with third parties for their direct marketing.

12. Consumer health data policy

This section is our consumer health data privacy policy under Washington’s My Health My Data Act, Nevada SB 370 and similar laws. Veram is not a health service, but the optional steadiness and closeness scores are inferences that could relate to a person’s mental health, so we treat them as consumer health data wherever you live.

• What: three scores from −2 to 2: emotional steadiness, a tendency to seek reassurance in closeness, and a tendency to keep distance. Each comes with a confidence. No quotes are stored with them.
• How we collect them: our AI models infer them from your profile and answers, and your own AI may refine them if you paste its scoring. Only after you switch them on in your deep profile; that switch is your consent, which you can withdraw at any time.
• Why: solely to avoid introductions that research suggests tend to hurt both people (for example, pairing someone who strongly seeks reassurance with someone who strongly keeps distance). They are used only when both people have switched them on.
• Who receives them: no other member ever sees them. They are processed by Anthropic PBC (to compute them) and stored with Supabase Inc., both acting as our processors. We do not sell them, share them with any other third party, or use them for advertising.
• Your rights: to confirm whether we hold them, to access them (they are shown to you in your deep profile), to delete them (switch them off, or delete your account), and to withdraw consent. Each takes effect immediately. You may appeal any refusal as described in section 10.

13. People in the EU and UK

Veram is designed for and offered primarily in the United States. If the GDPR or UK GDPR applies to you, our legal bases are listed in section 2; where we rely on legitimate interests, they are running a safe, secure and working service, and you may object. You also have the rights to restrict processing and to withdraw consent at any time without affecting earlier processing. You can complain to your data protection authority: in Spain, the Agencia Española de Protección de Datos (aepd.es); in the UK, the Information Commissioner’s Office (ico.org.uk). Where the law requires us to appoint a representative in the EU or UK, we will name them here.

14. Children

Veram is only for adults 18 and over. We ask for your full date of birth and check it on our servers. If it shows you are under 18, the account is closed to introductions straight away and its photos are deleted. We do not knowingly collect data from anyone under 18, and we delete any such account as soon as we learn of it. If you believe a minor is using Veram, report it to hello@veram.io.

15. Cookies and similar technologies

Strictly necessary: your sign-in session and the choices you make on this site (such as your cookie choice and progress through setup), kept in cookies and your browser’s storage. Analytics: one cookie, only if you accept in the banner; you can change your mind by clearing this site’s data in your browser or writing to us. Recordings of our public pages (section 2) set no cookie of their own. Where we ask about cookies they follow your choice in the banner. Elsewhere the switch below turns them off for this browser together with ad measurement, and a Global Privacy Control signal turns both off.

Page tests: where we don’t ask about cookies, a cookie named veram.ny keeps, for 30 days, which of two versions of our New York page you were shown, so it stays the same between visits. Where we ask (the EU, the UK and Switzerland) it is never set and everyone sees the same version.

Ad measurement: in the United States, the Meta pixel sets a cookie named _fbp, the TikTok pixel one named _ttp and the Google tag one named _gcl_au, which those companies use to recognise the browser that saw or clicked one of our ads. They never run in the EU, the UK or Switzerland, or when your browser sends a Global Privacy Control signal. You can turn them off for this browser here:

16. Changes

We will post any change here with a new effective date. For material changes we will email account holders before the change applies, and where the law requires it we will ask for your consent again.

17. Contact

Promvie, Inc. (Veram), 1111B S Governors Ave STE 28868, Dover, DE 19904, United States · hello@veram.io