This is a working draft, and no lawyer has read it.
There is no App Store listing, nothing has ever been sold, and nobody has agreed to anything — so none of this is in force. It is published early so that the terms are not written in a hurry the week before launch, and so that whoever reviews them can see what the app actually does.
Half of what a finished set of terms needs isn't drafted at all. §8 lists those headings with nothing under them, on purpose: a terms page that shows its holes is more use to the lawyer who has to fill them than boilerplate that hides them.
Questions about any of it: lapka.admin@gmail.com.
Written 9 August 2026, revised 12 August 2026 — §7 and §9 changed — and again on 13 August, when the app grew a box for telling us what to build: §4, §7, §8 and §9. Revised a third time on 5 September, when four sentences here were read against the code and found to be claiming more than it does: §3, §4, §5 and §6. A review of our launch markets has not started.
What's on this page
01
Nothing here is live #
There isn't yet an app to agree to terms about.
No listing, no subscriptions sold, nobody charged, no accounts in production, and no company. This is a draft of what the terms will say.
02
What Lapka is #
Behaviour training help for cats — teaching tricks, and undoing habits you'd rather not live with.
You describe your cat, Lapka builds a ladder of small steps, and what you report after each session changes what comes next.
03
What Lapka is not #
Lapka is not a vet and does not give veterinary or medical advice.
It doesn't examine your cat, it doesn't diagnose, and it never suggests treatment. If your cat is unwell, the answer is a vet, every time.
There is one place it touches health, and it is a handover rather than help. Three signs — breathing, back legs, not eating — each have one plain question attached, and only your answer to that question can stop the app. When it stops, it says your cat needs a vet and roughly how soon, gives you a sentence to say on the phone, and offers to find one near you. It tells you it cannot say what the thing is, and it is never behind an account or behind the price.
Its coverage is narrow, and you should know exactly how narrow. It is a short list of hand-written rules, not a model making a judgement. All three signs are opened only by matching English phrasing in text you typed, in an app that also ships in Ukrainian and Spanish. Somebody who records a voice note instead of typing opens nothing at all, and somebody who writes nothing gets no checks. It is not a safety net. It fires in a few specific situations and it will miss things.
This paragraph said something weaker until 13 August and the correction is worth printing rather than swapping in. It described four signs, said one of them was opened by a box you could tick rather than by English words, and called that one "the only door that works in any language". The fourth sign — the urinary one — was taken out of the app on 12 August, and that ticked box belonged to it. So the box went with the sign, there are three rather than four, and every remaining door now depends on you having typed English. This page was claiming a wider safety layer than the app has, which is the wrong direction to be wrong in about a cat that might need a vet.
And the reason a voice note opens nothing changed on 23 August, though the fact did not. This paragraph used to say the words came back only after you signed up, and that the checks had already run by then. Transcription moved in front of the account that day, so the words are on screen while you are still answering — and they still open nothing, because the recording and what we wrote down from it are kept as a separate thing from the paragraph, and these rules only ever read what you typed. Same gap, a different reason, and worth correcting rather than leaving as a sentence that happened to reach the right answer.
04
Your part #
You are the one in the room, and you know your cat.
The plan is built from what you tell us and Lapka cannot see your cat. Use your judgement, stop any session your cat finds too much — the app is built to treat that as information rather than as failure — and never read anything here as licence to punish or frighten a cat. Nothing in a Lapka plan will ever ask you to. If something reads that way, we have got it wrong and we would like to know.
Since 13 August there is somewhere to say that. On your cat's page, at the foot of a session, and — since 5 September — in Settings, there is a box for telling us what is broken or what should exist. It goes to the person building this and to nothing else — no model reads it, which is §6 of the privacy notice — and it does not reply. It is not a way to get help with your cat. If a plan has told you to do something that would hurt one, stop, and use lapka.admin@gmail.com so that a person can answer you.
05
Accounts #
One account per person, and one cat at a time.
The database enforces a single active cat for each account, which is not the same as one ever. Start fresh after a reinstall and the previous cat is retired rather than deleted — their intake, their programme, every report and their photo all stay on our server, out of the app's reach and out of yours, and only asking us takes them. This line said "one cat per account" until 5 September; the privacy notice corrected the same sentence on 15 August and this page did not follow, so §6 there is the account of what actually stays. Multi-cat households are not supported in the first version — that is a product limit rather than a household rule, and nothing in the app knows what a household is.
06
Money #
$2.99 a month, or $24.99 a year — which works out at $2.08 a month.
Three days free to start — on the monthly plan, and only there. A year bought outright begins paid. It is one free trial per person rather than one per plan, which is the store's rule and not ours, and this page printed it directly under both prices until 5 September as though it covered them both. The free days are the same paid plan with billing held back, not a free version, and there is no free tier: without a subscription there is nothing to use. It renews on its own. Billing, renewal and cancellation go through whichever store sold it — the App Store on an iPhone, Google Play on Android — and that store's rules govern refunds. iPhone is where this ships first, and nothing has been sold on either.
07
A model writes the plan #
Said plainly, because it should be.
When it is switched on, the deeper questions, the programme and the weekly check-ins are written by a large language model from what you told us. A model can be wrong. The app is built so that when it cannot answer something well it says so, rather than producing something that has the shape of an answer.
This changed on 12 August. Until then no model was switched on and a deterministic stand-in wrote those things instead; now the keys are set and real programmes have been written by a real model. A stand-in is still what runs where a key is absent, and everything one produces is marked as a stand-in in the payload — but it is the exception now, not the rule. Two vendors sit behind that one seam, Anthropic and OpenAI, and which of them writes your plan is a setting; the privacy notice names both for that reason.
What you write, say and report is yours, and none of it is ever used to train a model. Not the intake, not the paragraph you typed, not a report, not a recording, and — since 13 August — not a word of what you put in the box described in §4. That was an intention on this page three days ago and this page said so, because an intention is not a policy. It is a policy now, and a promise we are making rather than one we are hoping to keep.
The box goes further than that: nothing you write in it reaches a model at all, for any purpose. It is not summarised, not sorted and not turned into anything — it is read by a person, the way an email is.
A voice note exists only for as long as it takes to turn it into text, and is not kept afterwards. The privacy notice describes exactly where that is true today and where it is not yet — §3 there — because the phone has a step still owed, and a promise with a known gap in it should carry the gap next to it.
We do not sell any of it. What we still cannot tell you is the terms our providers handle it under: choosing providers whose terms we are happy to stand behind is part of the review that hasn't happened, and we would rather say that than make a promise on somebody else's behalf.
08
Everything a lawyer still has to write #
These have headings and nothing under them. That is deliberate.
Intellectual property
Acceptable use
Termination
Warranties and liability
Governing law and disputes
How we tell you these changed
09
Contact #
lapka.admin@gmail.com, and it goes to a person.
Added 12 August 2026, and it is the first address this project has had. No list, no automation, no ticket number — the person building Lapka reads it and answers it. Ask about anything on this page, tell us we have got something wrong, or ask us to delete what we hold, which is §10 of the privacy notice.
The box in the app is not this. It reaches the same person and it goes one way: no reply, no ticket, nothing to check back on. It is for what should be built and what is broken. Anything that needs an answer — your cat, your money, your data — belongs at the address above, and this paragraph exists so that nobody types something urgent into a box that will never write back.