Effective date: September 8, 2026
Last updated: October 20, 2026

1. What this covers

This policy describes what Launchism collects, why, who can see it, how long it's kept, and what you can do about it. It's written to be read, not skimmed past — if anything here is unclear, that's on us to fix, not you to decode.

The short version of what Launchism is: a coaching and companionship product for people navigating job loss — not therapy, not a medical device, not a crisis service — designed so that the moment you need more than it can give, it hands you to real human help, for free, without ever having monetized the moment that got you there. The full description of what that means in practice is in the Terms of Service.

2. What we collect

• What you tell us at signup and in daily check-ins: how long you've been out of work, your CV/role/company/industry, where you are in your search, and — once you choose to share it — more personal information: the reason your job ended, your family situation and financial runway (as ranges, not exact figures), how you're feeling, and, directly, whether you're having thoughts of harming yourself.
• What you say in conversation with Launchism, including anything you disclose about your emotional state, relationships, or situation.
• What the product infers from that — patterns it notices across weeks (for example, "this keeps coming up"), used to make its coaching more useful to you.
• Anything you choose to import — a CV, a spreadsheet of applications, or similar.
• Ordinary account and usage information — that you signed up, when you used the product, and (once payment exists) that you paid or cancelled.

3. How we use it

• To have the coaching conversation itself, and to remember your situation across sessions so you don't have to re-explain it.
• To notice patterns in what you've told us over time, so the coaching adapts to you specifically.
• To detect and respond to safety concerns — see Section 6.
• To run the product: keep your account working, respond to support requests, and process payment once a paid tier exists.
• To meet legal obligations, including California's SB 243, which will require us to report annually (starting July 2027) an aggregate count of how many crisis referrals we issue — a number only, never your identity or content — and to defend against a payment dispute using only account and billing metadata (see Section 8) — never your conversation content.

What we never do with it: your emotional or situation data is never used for advertising, never sold or shared with data brokers, and never repurposed under an "anonymized" label as a workaround. This isn't just policy — it's enforced in how the product is built, not left to a promise alone.

4. Your consent

Using Launchism beyond the free preview requires one consent, given once, covering everything in this policy together — there is no version where you consent to some of this and not the rest. The consent screen states it in these words:

Withdrawing consent means deleting your account — stated on the same screen as: "Withdrawing this means deleting your account." There is no partial-withdrawal mode. If you decline consent at signup, you're offered a choice: leave and keep your account as-is (your answers so far are retained on the normal 90-day schedule described below, and you can come back and consent later), or leave and delete everything gathered so far immediately. Nothing is deleted automatically just because you declined — deleting is something you choose.

5. You're talking to an AI

Launchism tells you plainly, before you can interact with it, and as a standing line under the input afterward: "I'm an AI, not a person." This is not buried in this policy — it's in the product itself, because it matters in the moment, not just on paper.

6. Crisis detection, and what happens when it activates

The automated system — separate from the coaching conversation itself — checks what you write for signs of self-harm, an intent or plan, hopelessness with risk markers, harm toward another person, or an acute crisis. It's deliberately tuned to flag even when it isn't certain, because missing a real crisis matters more than a false alarm. It can also miss a real one, or respond late — it's automated, not a guarantee. If you're in immediate danger, please contact emergency services directly rather than relying on this system to notice.

When it flags something: Launchism checks in with you directly, and — regardless of how you answer — shows you a real, human-staffed resource list (in the US: 988 and the Crisis Text Line; everywhere else: the worldwide directory at findahelpline.com; always: local emergency numbers for immediate danger). This never requires a login, is never limited by your payment status, and is always available at /help even if you're not signed in.

What that page does and doesn't do: it doesn't check who you are, and nothing you read or tap on it is tracked by us. It is *not* invisible to everyone — like any page on the internet, the request to load it passes through our hosting provider's ordinary infrastructure and may appear in their standard request logs. We don't read those logs to find out who visited, and nothing about your visit is recorded in your Launchism account or conversation history.

When a human reads your words, and when they don't. A flag from the automated system is what leads the founder to read conversation content — specifically the flagged message and enough of the surrounding conversation to understand the situation, not your full history by default. Ordinary account support (billing questions, login trouble) never involves reading your conversation content — only account-level information like whether you're signed up or subscribed. This is logged internally as an access event, but you will not receive a real-time notification each time it happens. There's no guaranteed response time, and Launchism doesn't collect your phone number or address — so even after a human reads what you wrote, reaching you outside the product may not be possible on that alone.

We do not promise absolute confidentiality. If what you write leads a reasonable person to believe there's an imminent threat of serious physical harm to you or to an identifiable other person, the founder may — after reviewing it himself, not automatically — contact emergency services or law enforcement.

If you think a safety flag on your account was handled badly, or missed something, tell us — see Section 15 for how to reach us.

7. Retention — how long we keep things

We keep your conversation, situation, and emotional-disclosure data for 90 days after your last activity, then it's automatically removed. The minimal record of a crisis flag (when it happened, its severity tier, what happened next — not a transcript) currently rides this same 90-day window as our working default, and is used only to avoid over-messaging you, for internal safety review, and for the aggregate legal report in Section 3 — never for anything else.

(Note on the alpha specifically: the automated 90-day purge is built but not yet live within the alpha's timeframe — manual deletion, described below, is the working path for now.)

8. Sharing your information

We share information only as needed to run the product, never to monetize it:

• The AI model that powers coaching, the separate safety-classifier system, and the database host that stores your account's information each process only what's needed to do their specific job, and none may use your data to train its models or for any other purpose. How long each one keeps your words is stated for each below.
• Model provider: Anthropic, PBC, the company that makes Claude, receives each message and writes the reply. Anthropic does not train its models on your words. Under its commercial terms it deletes each request and reply within 30 days; if its automated safety systems flag a request, it may keep that request for up to 2 years. We have asked Anthropic for an arrangement under which it stores nothing; until it agrees, these terms apply.
• Backup provider: if Anthropic can't answer at that moment, the same message goes to OpenRouter, Inc., which forwards it to Relace, the company whose servers run DeepSeek V4.1 Flash, an open-weights model. We pin every backup request to Relace, so no other company on OpenRouter's network receives your words. Each backup request carries OpenRouter's zero-data-retention flag: OpenRouter sends it only to an endpoint that stores nothing, and refuses the request rather than send it anywhere else.
• Safety-classifier provider: the same companies, in the same order: Anthropic first, then OpenRouter and Relace as the backup, under the same terms.
• Reply check: before a reply is shown to you, a second model reads it against what you have told the app and corrects anything that does not match. For this, your recent conversation, the draft reply, and what you have told the app (your sign-up answers, your CV, what you saved, your plan, and the messages you have sent, with their dates) go to the same companies, in the same order and under the same terms.
• Database host: pending — not yet under contract. This section will name the vendor once a signed agreement is in place.
• Payment processor (once a paid tier exists): Stripe, for billing and subscription management.
• Payment-dispute processing: if you dispute a charge, we may share limited account metadata (login history, session counts, feature use, billing correspondence) with our payment processor to respond — never conversation content, emotional disclosures, or crisis-flag information, without exception.
• Emergency and safety exceptions: see Section 6.
• Legal requirement: if we're required by law to disclose something, we will, and we'll tell you unless the law prohibits it.

We do not sell your information, and we do not share it for advertising purposes, under any circumstance.

9. Your rights and controls

• Download your data. In Settings, choose "Download my data". You get a file with everything you've told Launchism — your check-ins, remarks, conversation content, and anything you imported. (This does not include Launchism's internal pattern insights, since those are derived, not something you gave us.)
• Delete your account. In Settings, choose "Delete my account". You are offered a copy of your data first. Deleting removes your situation record, conversation history, insights, and activity log at once; backup copies are purged within 30 days. Deletion is never delayed or interrupted by an active safety conversation. If you can't sign in, email hello@getlaunchism.com from the address you sign in with and we'll do either for you within 30 days.
• What deletion doesn't remove: the bare fact that an account existed and, once payment exists, that it paid or cancelled — a business record that never contained anything you told the product about how you're doing.

10. Where this applies

Launchism is available globally from day one. If you're in the US, the crisis resources above are written for you directly; if you're elsewhere, the worldwide directory at findahelpline.com is provided because we haven't yet verified country-specific numbers against their operators — we'd rather send you somewhere reliable than guess.

No EU (GDPR Art. 27) or UK representative has been appointed as of this policy's last update. This gap is being closed with counsel before wider launch.

If you're a California resident, state law (the CCPA/CPRA) gives you the right to know what we collect about you, to delete it, to correct it, and to not be discriminated against for exercising any of these rights — all of which you can already do through the download and delete controls in Section 9. Some of what you tell Launchism about your health and emotional state is treated as sensitive personal information under that law; consistent with everything else in this policy, we don't sell it and don't use it for cross-context advertising. (This paragraph states our practice; whether Launchism meets the statutory size/revenue thresholds that trigger CCPA's formal notice requirements hasn't been confirmed with counsel yet.)

11. Consumer health data (Washington and similar laws)

Because information about your emotional and mental state may count as "consumer health data" under laws like Washington's My Health My Data Act, that data gets its own, separate policy — see the Washington Consumer Health Data Policy — in addition to everything above.

12. Age

Launchism is for adults, 18 and older. We don't currently verify age at signup. If it becomes clear an account belongs to someone under 18, Launchism stops processing new information, says plainly that it's built for adults, and gives that person the same export-or-delete choice described in Section 9 — the same as any user.

13. Security

We take reasonable technical and organizational steps to protect your information, but no system is perfectly secure, and we can't promise otherwise. If a breach affecting your information occurs, we will tell affected users what happened without unreasonable delay.

14. Changes to this policy

If we change what we collect, how we use it, or who can see it, we'll tell you before the change takes effect — not just update the date at the top of this page.

15. Contact

Launchism is operated by Tzvika Besor. You can reach us by mail at: 62 Krinitzi St., Ramat Gan, Israel.

16. The waiting list and our website

If you join the Launchism waiting list at getlaunchism.com, we keep your email address and the time you joined. If you choose to answer, we also keep when you were laid off, what kind of work you did, and whether you're open to a short conversation. We use this only to tell you when Launchism is live and, if you said yes to a conversation, to ask for one. We don't sell it, and we don't share it for advertising. It's stored in a database hosted by Neon, separate from the product's own data, and the website runs on Vercel. To stop abuse, we keep a scrambled (hashed) form of your internet address for up to one day. To know whether people visit, the site counts page loads: the day, whether you arrived from a search engine or another site, and whether you're on a phone or a computer. It also uses Vercel Web Analytics, which counts visits without cookies and without identifying you. To come off the list, reply to any email from us or write to hello@getlaunchism.com, and we'll delete your entry.