AppliAid is operated by Trophe Labs LLC, a California limited liability company (“we,” “us”). Trophe Labs decides what AppliAid collects and why, and is responsible for your data under this policy. The companies listed under “Where it lives” process it for us and only on our instructions. How to reach us is at the end of this page.
Your email address (for sign-in, and for the emails we send about your account and your subscription), the career records you import or state (positions, accomplishments, projects, education, publications, skills), the job postings you paste, the fit analyses and documents generated for you, and your answers to intake questions. That is the product: a career record you control.
We also count how the product is used: a public page viewed, a sign-in completed, a CV parsed, a fit run, a document generated. Those records hold numbers, timings, outcomes, the domain of the site that linked you here and the campaign label on our own links, whether the visit came from a phone, and the domain of your email address (the part after the @, so we can tell a university from a webmail account). They never hold text you wrote, and they never hold your email address itself. If you would rather not be counted, there is a setting for that under Settings: it stops new counts against your account and removes the ones already stored. A visit before you sign in is counted under a cookie that nothing can tie to you until you do, so that part cannot be honoured in advance, and we would rather say so.
When you delete or edit a record, on the review screen or on the graph page, we keep a note of the shape of that correction and nothing else: what kind of record it was, whether it came from your CV or from you, roughly how long it had been there, and for an edit, the name of the field you changed (a title, a date, a proficiency), never what it said before or after. When AppliAid reads a CV and meets something it has no kind of record for, a grant or a committee seat, say, it keeps a note of that too: the word for what kind of thing it was, and not the thing. None of these notes hold text from your CV, an identifier for the record, or anything that links them to you or your account. They are counts, not copies, kept so that we can measure which kinds of record AppliAid reads badly and which it cannot read at all. If you would rather we did not keep them, there is a setting for that under Settings, and it applies from your next correction.
There is one thing we keep that does hold text from your CV, and only if you switch it on. Under Settings, accounts opened from 12 September 2026 can turn on Keep my corrections as examples. It is off unless you turn it on, and every part of AppliAid works the same either way. With it on, each time you correct something AppliAid read from your CV we keep one example: the line it read, what it thought that line said, and what you changed it to. From your next import we also keep that line for records you have not corrected, because the line is gone by the time you correct something and the example needs it. These examples are tied to your account, listed for you under Settings, and you can delete any of them.
We use them for two things: checking that AppliAid still reads CVs correctly before we change how it reads them, and as worked examples inside the instructions we give the model. We do not train or fine-tune a model on them, and if we ever want to we will ask you again rather than treat this as permission. A line that looks like it touches health, religion, politics, union membership, ethnicity or sexuality is never kept, whether the switch is on or not. Turning the switch off deletes every example and every stored line.
Fourteen days after a fit run, the posting’s page asks two questions with fixed answers: did you apply, and did you hear back. Your answers are kept with that fit run, and they are the only way to find out whether a score predicts anything. Each run asks once; skipping it is final for that run, and there is a setting under Settings that stops the questions on every run. An answer you have given can be removed on the posting’s page.
Your CV file is read once to extract records. You review what came out, and only what you approve is saved. The file is deleted when the extraction finishes. Between the reading and your approval, the draft records wait in your account. If you discard the import, they go with it.
Two things can rest on our servers before you have an account: a CV, and a job posting. If you use the box on the home page — attaching a CV, pasting a posting, or both — what you left is held against a one-time session identifier, so your first fit can run the moment you arrive. Held, not read: nothing is parsed and nobody looks at it until you sign in and ask for the run. If you sign in, a CV is read and then deleted like any other import, and a posting waits in your account until you press the button that judges it. If you do not sign in, everything you left is deleted automatically within 24 hours, and no account is created.
Different things have different lifetimes. This is all of them.
Data is stored in a Supabase (Postgres) database in the United States, with row-level security scoping every record to your account. The app runs on Vercel. Six other processors touch limited data to make the product work: Anthropic runs the language models that parse your documents and generate output (your graph records and pasted postings are sent for those runs; for interview prep, the company and role name are used in live web searches); Resend delivers sign-in emails and the emails we send about your account and subscription; Sentry receives error reports when something breaks so we can fix it; Deepgram turns a spoken answer to an intake question into text, if you choose to speak one instead of typing it; Stripe takes payment if you subscribe, receiving your email address at checkout and holding your card details, which never reach us; and Cloudflare runs a bot check on the sign-in form, which sends your network address and some details of your browser to Cloudflare so that a machine cannot feed other people’s addresses into the form. Cloudflare says the check processes only what that decision needs.
Like every web service, the hosting layer keeps ordinary server logs (IP address, request details) for a short time for security and debugging. If you use AppliAid from outside the United States, your data is still stored and processed in the United States.
Your data is not sold, rented, or shared with advertisers or data brokers. There are no ad trackers or third-party analytics scripts on the site. No one but you sees your graph, your fit reports, or your documents — with the two exceptions any honest policy has to name: a valid legal order could compel disclosure, and if AppliAid were ever acquired, your data would move with the service under this same policy and you would be told first.
Nothing you put into AppliAid is used to train models. AppliAid has no models of its own. Anthropic, which runs the models that read your CV and write your documents, does not train on what we send it. Deepgram, which turns a spoken answer into text, would by default, and we tell it not to on every single request.
Four, all first-party, none of them shared with anyone.
The sign-in cookie that keeps you signed in.
While you are signed in, a cookie holding today’s date and nothing else, so we record that your account was opened today once rather than on every page. It expires after a day. The date it records is what keeps your account out of the 12-month deletion described above.
An anonymous cookie set on your first visit, holding a random identifier and nothing else, so we can count how the site is used, and which site or campaign link sent you here. When you sign in, that identifier is joined to your account.
If you leave a CV or a posting in the box on the home page before signing in, a short-lived cookie holding the one-time identifier for what you left, so it finds you again when you sign in. It expires with what you left, within 24 hours.
The bot check on the sign-in form runs in a frame served by Cloudflare. Anything it keeps lives in that frame under Cloudflare’s terms; nothing from it is set on appliaid.com.
There are no marketing cookies, no ad trackers, and no third-party analytics scripts.
You can edit or delete any record on the graph page, and deletions are real deletions, not soft flags. You can delete your whole account, and everything in it, from Settings: it goes at once, a subscription is cancelled first so nothing is charged after, and there is no undo. Everything else works by email from your sign-in address to support@appliaid.com: ask for a copy of everything AppliAid holds about you, ask for a correction, or ask us to delete the account for you — done within 30 days, usually much sooner. These rights are yours wherever you live. How long each kind of thing lives is set out above.
If you left a CV or a posting on the home page and then changed your mind, do not sign in. Everything you left deletes itself within 24 hours and no account is created. There is no button for this, and we would rather say so than pretend otherwise: until you sign in we have no way to know which held session is yours, so we cannot act on a request about it. Signing in makes it yours, and then everything above applies.
AppliAid is not directed to children and requires users to be at least 16. We ask before a file is attached, not after an account exists. If we learn an account belongs to someone younger, we will delete the account and its data.
If this policy changes in a way that matters, the change will be described here and the date above updated before it takes effect.
Requests about your data, and questions about this policy, go to support@appliaid.com, from your sign-in address if the request is about your account. By post: Trophe Labs LLC, Attn: Privacy, 2108 N St, Ste N, Sacramento, CA 95816, United States.