Privacy Policy
Last updated October 12, 2026.
Tellwell is in pilot testing. You hand us your employment history, which is about the most personal document most people own, so this page says plainly what happens to it. If anything here is unclear, ask us at tellwell.support@forhumankind.io.
What we collect
The resume you upload, the job posting you point us at, and anything you type during a session. If you create an account, your email address and your subscription status. Plus the ordinary technical records any website keeps — pages visited, rough location from your IP address, browser type, and errors.
What we never do
- We don't sell your data, and we don't share it for advertising.
- We don't let AI providers train on your content. Every provider we send it to is configured, or bound by its terms, not to train on it. Some may keep a request for a short period to detect abuse, and the list below says which.
- We don't record your screen. Session recording is switched off, and the analytics we do run is configured to mask text so your resume can't leave the page that way.
- Your resume text never appears in an analytics event.
How your session is used
Your resume and the job posting are sent to a large language model to produce a draft resume and cover letter, and to a smaller set of intermediate steps along the way (pulling out your experience, deciding what to emphasise, checking the draft). We also keep a record of each step so we can work out why a bad result was bad.
To recognise the same skill however a posting words it, and to see which of your experience a posting is asking about, we also turn short pieces of text into a numeric fingerprint (an “embedding”) and compare it with the European Commission's public ESCO skills list. The text can be a line from the job posting and, for some features, a line from your resume. The fingerprint is used for that one comparison and then discarded. What we keep is which skills matched, not the fingerprint. This is done by an embedding model, which is a different model from the ones that read and write, so it can be a different provider.
Beyond the readings themselves, we keep a few things worked out from your resumes, your answers and the postings you read. They are about you. We don't share them beyond the providers above, which do the reading, and we don't compare them with other people's:
- Which skills from the ESCO list each posting asked for, and which requirements nothing in your resume or answers supported.
- The skills that keep showing up across the postings you read, and how you responded to them (“I hadn't thought of that”, “that's not me”). Your response is stored as the skill's ESCO identifier and nothing else.
- Roles we suggested as somewhere you might be headed, and the ones you said no to, so we don't suggest them again. Also stored as the role's ESCO identifier only.
- For roles you choose as goals: the route to each, which skills your own words show, imply, or don't touch, the lines of your resume and answers it rests on, and a few suggested first steps. A goal says what you want, which makes it the most sensitive of these, so we say so here.
Sessions are deleted on the schedule below, but a goal's route quotes lines of your resume and so is kept after the session it came from is gone. Setting a goal down hides it; removing it deletes it and its route. While you're signed in you can download a copy of your readings, answers and the things worked out from them at your data export, and anything else you can ask us for by email. Deleting your account deletes all of it.
Who we share it with
We use these providers to run the service. Each one only receives what it needs, and each is bound to handle it on our instructions.
- OpenRouter — routes our AI requests to model hosts, currently Z.ai's GLM models for reading and writing and OpenAI's embedding models for matching. Receives your resume, the job posting, and the short pieces of text described above. Requests are restricted to hosts that don't retain or train on the content, which may be located outside your country, including in the US and China.
- OpenAI — may receive the same short pieces of text directly, for its embedding models, when we call it ourselves instead of through OpenRouter. We use OpenAI's API terms, under which API content isn't used to train its models; it may keep requests for a short period to detect abuse. Servers are in the US.
- Different models do different jobs, and we change which provider does which as we find what works best and costs least. When we add a provider or move a kind of text to a different one, this page changes first.
- Supabase — database, file storage and sign-in. Holds your session data and email address.
- Vercel — hosting. Handles every request and keeps short-lived server logs.
- Stripe — payments. Receives your billing details directly; we never see your card number.
- PostHog — product analytics and error reports, and only with your consent. Receives page views and event names, never document content.
Cookies
Two kinds, and only one of them is a choice. The cookies that keep you signed in and carry you through checkout are strictly necessary — without them the product doesn't work, so they don't need consent and there's no way to turn them off and still use Tellwell. The analytics cookie is optional. Decline it and we fall back to counting visits without a persistent identifier.
You haven't chosen yet — the banner at the bottom of the screen is waiting for you.
How long we keep it
Sessions are deleted after 12 months without activity. If you ask us to delete your account we act on it within 30 days, apart from what we have to keep for tax and accounting — Stripe payment records, which are kept for six years and contain no resume content. Analytics events are kept for 12 months. Your saved answers, your Career Story summaries and the goals and other things worked out from your sessions are not on that schedule: they stay until you ask us to delete them, or delete your account.
Security
Traffic is encrypted in transit and data is encrypted at rest. Sessions are protected by database row-level security, so one person's session can't be read by another even if something else goes wrong. Card details never touch our servers. Access to production data is limited to those who need it. No system is perfect, and if we ever have a breach affecting you we'll tell you.
Your rights
You can ask for a copy of your data, ask us to correct it, ask us to delete it, or object to how we're using it. Email tellwell.support@forhumankind.io and we'll respond within 30 days. If you're in the UK or EU and you think we've got it wrong, you can also complain to your national data protection authority.
Contact
Tellwell is operated by For Human Kind. Reach us at tellwell.support@forhumankind.io.