Privacy

This says what we hold, why we hold it, who else sees it and how to get rid of it. One thing up front, because it is the question worth asking of anything that reads your resume: we do not train AI models on your content. Not in any form — not anonymised, not aggregated, not “to improve our services”. What you give us is used to do the thing you asked for, and for nothing else.

1. Who holds your data

Possena is operated by Project 66 Inc., a Delaware corporation (file number 10766086) with its business address at 16192 Coastal Highway, Lewes, Delaware 19958, United States. Because we operate from Quebec, we are subject to Quebec’s Act respecting the protection of personal information in the private sector (as amended by Law 25) and to Canada’s PIPEDA. If you live in the United States, the United Kingdom or the European Union, your own law gives you rights too, and we have not written you a separate section: the rights in §7 are given to everyone who uses Possena, wherever they live, because keeping two standards and applying the lower one to some people is not something we were willing to build.

Our Privacy Officer is Olia Stasiuk, reachable at hello@project66inc.com or at the address above. Write there for anything in this document — access, correction, deletion, portability, or a complaint. We answer within 30 days.

2. What we hold, and why

The documents you upload, kept exactly as you sent them. The facts read out of them, each tied to the passage that states it. Job postings you add or link to. Search intent — the roles, places and conditions you tell us about. Transcripts of voice notes you choose to send. Your account: an email address, and a payment reference if you buy something. We hold each of these to do the thing you asked for, and for nothing else.

Two records exist that you did not type, and it is easier to say so than to let you find them in the export. One is a short log of the steps taken in the product — a visit, an upload, a purchase — kept so we can see where the product loses people. The other is a line for every AI call made for you: which model, how many tokens it used, how long it took, what it cost. Neither one holds any of your text. Not the prompt, not the answer, not a sentence of your resume — the columns for those do not exist. Delete your account and the step log keeps its rows with your name taken off them, in the same way and for the same reason as the credit ledger in §6.

3. Before you have an account

You can start without one. The first thing you do begins an anonymous session, so an upload survives a refresh and belongs to you; signing up later claims it. If you never come back, that session and everything in it expires after fourteen days and is removed.

4. Your voice

Audio goes from your microphone straight to the speech-to-text service over a link this server signs and never sees again. No recording of your voice reaches our servers or our storage. The transcript is the only thing that does, and only when you send it. Audio, transcript and confirmed fact are three separate things.

5. Who else sees it

We use a small number of services to run the product, and text you give us may pass through them:

• Supabase — the database, sign-in, and the storage your original documents sit in. Sign-in is also what sends you email: the code that gets you in, the link that resets a password. Supabase is given your address and the message, and nothing else goes out by email at all — there is no newsletter and no marketing list.
• Amazon Web Services (Bedrock) — the models that read a posting and write an application. Your resume text and the posting reach them. The models are Anthropic’s and OpenAI’s, but they run inside Amazon’s service: what you give us does not go to either company’s own systems.
• Amazon Transcribe — speech to text, straight from your browser.
• Netlify — hosting, and therefore the request logs any web server keeps.
• Stripe — payments. Your email address goes to Stripe so that a receipt reaches you and so that the same person is not made twice. Your card number and your billing address go to Stripe and never to us: they are typed on Stripe’s own page, not ours. What we keep is a reference to your record at Stripe. When your account page shows a card, it shows the brand, the last four digits and the expiry, read from Stripe at that moment and stored nowhere here.
• Cloudflare — the check that tells a person from a script, on the pages you can use before you have an account. It is given your IP address, the token its own widget produces, and ordinary signals about your browser — enough to answer that one question and nothing else. It is never given your name, your documents, or anything you have written. §8 says more, because it is the one piece of code on this site that is not ours.
• Your browser’s notification service — Google’s, Apple’s, Microsoft’s or Mozilla’s, whichever your browser uses — but only if you turn notifications on, and it cannot read what we send. §9d says how that works.
• Discord — only if you choose to use our bot there. §9b says what that means, and it is different from the rest of this list.

Your material is not sold, and it is not used to train anybody’s model.

6. How long we keep it

For as long as you have an account, because that is what makes the work yours to come back to. Ask for the account to be deleted and career and voice content is erased after a twenty-four hour window in which you can change your mind. Records of payments and credits are kept for audit against your account reference, because a ledger that can be edited is not a ledger.

7. What you can do

Export everything the account holds, at any time, from your account page. Delete the account, with the window above. Mark any single fact private, and it stays out of everything written for you. Turn notifications off, in the browser that asked for them, whenever you like. There is no cookie banner to change your mind at, because there is nothing here to consent to — §8 lists every cookie this site sets and they are all doing a job you asked for. We do not sell or share personal information, so there is nothing to opt out of — asking is still free.

None of this costs anything, none of it needs a reason, and using any of it does not make the rest of the service worse for you.

8. Cookies

There are no advertising cookies on this site, and nothing here follows you from one website to the next. The cookies it sets are the ones it needs:

• Your session — set by sign-in, so the site knows it is you.
• ca_anon — the anonymous session described in §3, so work done before you sign up is still yours afterwards. Fourteen days.
• ca_composer — what you typed into the box on the front page, held for the screen that asks you about it; long words continue in ca_composer_1 to ca_composer_3. Fourteen days.
• ca_start — the job you brought from the front page, and where your first analysis is, so the steps after you join can pick them up. Fourteen days.
• ca_ref — a referral marker, if you arrived from someone’s invitation, so the credit reaches them.
• ca_src — which announcement brought you here, if you arrived from a link that said so, so we can tell which ones are worth doing again. It holds that word, and the campaign name if the link carried one, and nothing about you; the first one wins and is never overwritten, and it is not readable by any script on the page. Ninety days.
• ca_continue — keeps an opaque reference to what you were doing so Possena can return you to it after sign-in. It contains no resume, job, payment or provider data. It expires after 14 days and is cleared when the continuation is completed or is no longer valid.

Clearing your browser’s cookies for this site removes all of them. Doing that while signed out also drops any anonymous session, and the work in it goes with it.

One thing is kept in the browser rather than in a cookie, and only while that tab is open: which of the notices that something could not be read you chose to continue without while setting up, so it is not shown again on the next screen. Nothing that belongs to your career is ever kept there — not a fact, not a document, not a credit, not a score; the rule is enforced by a test that fails the build if anyone adds one. It is wiped the moment you sign out, so a shared computer does not hand your answers to whoever sits down next.

One piece of code here is not ours. On the pages you can use without an account — the box on the front page, a job sent in from your phone, the microphone — Cloudflare’s Turnstile runs in your browser to tell a person from a script before a session is started. Starting one is free and unlimited, which makes it the one thing a script could do all day, and this is what stops that. It is loaded from challenges.cloudflare.com; it receives your IP address and ordinary signals about your browser; and it may set a cookie of Cloudflare’s own so that a browser which has already passed is not asked again. It is a bot check and not an advertising tracker: Cloudflare is not given your name, your documents or anything you have written, it does not build a profile of you, and nothing it sees is used to show you an advertisement. Cloudflare is in the United States, so this is a transfer outside Quebec like the others in §11. There is no analytics script, no advertising script and no social button on this site; that is the whole of the third-party code.

9. Children

This is a service for people looking for work and it is not intended for children. We do not knowingly hold data about anyone under sixteen.

9b. If you send us something through Discord

You can send a job, or your resume, to our Discord bot. When you do, Discord handles that message before we ever see it, and a file you give to Discord is a file Discord holds — under their privacy policy, not ours, and we cannot make them forget it. We delete our copy when you ask; we have no power over theirs. If that matters to you, upload on possena.com instead, where deleting it deletes it.

When you use our Discord command we also keep what Discord tells us about you: your handle, the name and picture you have chosen to show publicly, any nickname you use in that server, your language setting, and which server you came from. We keep it so that a person who comes back is recognised rather than treated as a stranger, and so we can see which communities the product is useful in. It is public profile information, never message content — we do not have permission to read messages and have not asked for it.

9c. The Possena browser extension

There is a Possena extension for Chrome. It reads the job posting on the page you are already looking at, so a posting that our own server is refused can still reach your Job Match. It does nothing at all until you press its button. There is no script that runs on pages in the background, no timer and no list of sites it watches.

• What it reads, when you press it: the visible text of that one page and its web address. Not your cookies, not your browsing history, not what you have typed into a form, and nothing on any other tab.
• What it sends: that text and that address, to Possena, signed in as you — the same as pasting the posting in yourself. It can reach no other website. It carries no key of its own and stores no password.
• What it keeps: nothing. It stores nothing on your computer — no copy of the page, no record of what you captured, no settings.
• If you ask it to fill in an application: it types contact details you have already given us — your name, email, phone, city and the links you added — into the form in front of you. It asks us for those and for nothing else: no facts, no resume, no scores. It never presses Submit, it never writes over a box you have already filled, and it leaves alone every question that decides an application — work authorisation, sponsorship, visas, licences, salary, notice period — and every voluntary question about who you are, including gender, race and ethnicity, veteran status and disability. Those are yours to answer, and a blank is better than our guess.

Removing the extension removes everything it can do. Anything it already captured is in your Job Match, where you can delete it like anything else.

9d. Notifications

Possena can tell you about a job without you having to open it. That is off until you say yes, we never ask on your first visit, and saying no is the end of it — there is no second prompt and nothing works worse afterwards.

Say yes and your browser gives us an address to deliver to, which we keep with two keys it also gives us. That address belongs to whoever makes your browser — Google for Chrome and Android, Apple for Safari, Microsoft for Edge, Mozilla for Firefox — and our message goes through them, because that is the only way a notification reaches a phone that is asleep. They cannot read it. Each one is encrypted here, with those keys, before it leaves, and only your device can open it; what the delivery service learns is that something arrived for a browser, not what it said. It carries a job and never anything else — never a feature we have shipped, never a nudge to come back.

Turn it off in your browser’s own settings for this site, and the address stops working. When the delivery service tells us an address has been forgotten, we delete the row rather than retry it — partly because it will never work again, and partly because a stale address is how somebody else’s phone ends up buzzing.

10. Automated decisions

Possena scores how well a job matches what you have told us about yourself, and writes documents from your facts. Those are done by software and by AI models, without a person reading them first. None of it decides anything about you. It does not decide whether you get a job, whether you may apply, or what you are capable of — no employer receives a score from us, and no score is shared with anyone. Every score can be opened to see exactly which requirement it came from and which of your facts answered it, and you can tell us we read something wrong. If you would rather a person looked at something, write to the Privacy Officer above.

11. Where your data is

Our database, file storage, sign-in and sign-in email are Supabase; the site and its background work run on Netlify; the check that tells a person from a script is Cloudflare’s; payments run through Stripe; the AI models run on Amazon Bedrock; speech is turned into text by Amazon Transcribe; and a notification, if you have asked for any, goes out through whichever service your own browser uses. Sign-in can also go through Google, GitHub, LinkedIn or Discord if you pick one of those instead of an email code. These providers hold or process data outside Quebec, principally in the United States. They act on our instructions, are bound by contract to protect it, and none of them is permitted to use your content for their own purposes or to train their models on it.

11b. How it is kept safe

The whole of it travels over an encrypted connection, and your documents sit in private storage under your own account’s folder — there is no public link to any of them, and one is fetched only for the person it belongs to. Underneath that, the database itself decides who may read a row, account by account, rather than the application asking it nicely: a query run as you can only ever return your rows, which is why the export in §7 cannot accidentally contain somebody else’s. The keys that could read more than that live on the server and are never sent to a browser. The session cookie cannot be read by any script on the page. And the two records that must not be quietly rewritten — the credit ledger and the log of AI calls — are held by the database against edits and deletes, not by our promise not to make them.

None of that is a guarantee, and we would rather not pretend otherwise: no service can promise it will never be broken into. What we can say is what happens if it is, and §13 says it.

11c. When somebody official asks

A court, a regulator or the police may one day ask us for information about a person. If that happens: we ask to see the legal instrument that compels it, we hand over only what is actually demanded rather than everything about you, and we tell you it happened — unless we are forbidden by law from telling you, in which case we say so as soon as we are allowed to. We do not hand your material to anyone, official or otherwise, simply because they asked politely.

12. What we do not do

We do not sell your data, and we never will. We do not share it with employers, recruiters or advertisers. We do not use it to train AI models. We do not build a profile of you for anyone but you. We run no advertising on this site and carry no advertising trackers. The one piece of code here that is not ours is Cloudflare’s bot check, described in §8: it is there to keep scripts out, not to watch you.

13. If something goes wrong

If personal information we hold is lost or exposed in a way that poses a risk of serious harm, we will tell you and the Commission d’accès à l’information du Québec, promptly, and say plainly what happened and what to do about it. You can complain to the Commission, or to the Office of the Privacy Commissioner of Canada, at any time — you do not have to come to us first, though we would like the chance.

14. Changes

If we change how we handle your data in a way that affects you, we will tell you by email before it takes effect, and the date below will change.

Last updated: 14 September 2026.