Software built to bring a child back to it has a shape, and you will know it: points, coins, streaks, a class leaderboard on the whiteboard. Pomble has none of it. If that shape suits your school, there is plenty of it about. If it does not, there is not much else on the shelf.

Pomble is the other thing. Thinking games and a drawing Studio, and nothing that scores a child, ranks them against the table next to them, interrupts them, advertises to them or sells to them. No chat. No timers counting down. A child who finds a puzzle hard is not told they are behind, because there is nothing to be behind on.

What we are not claiming: that this makes children cleverer, calmer or better at maths. We have not measured it, and we would be sceptical of anyone who told you they had.

Where it would fit is narrow, and we would rather say so than have you find out in week two. Wet play. Golden time. The twenty minutes after an assessment. A small group who need something that is not more of the thing they have just struggled with. It is a library, not a scheme of work. It does not cover a curriculum and it is not going to.

Setting it up: one spreadsheet. First name, last name, class. Not date of birth, not UPN — we do not want them and there is nothing useful we could do with them. Before anything was saved you would see a preview of exactly what was about to happen, and a child missing from a re-import would never be quietly deleted; somebody would be asked. How long that takes in practice, we do not know, because nobody has ever done it. That is one of the things a first school would tell us.

Getting children onto it: an infant class would use a class code and a grid of names — tap your own name, no username, no password. A child picking their own name from a list already exists in the family version, on a single device; the class code, the roster and anything spanning two devices do not exist at all. One trade-off we will not talk around: a printed name card is a key, and whoever is holding it is that child. That is the reason older year groups would get passwords.

Week to week: three screens, and then we stop. Who has not started. A class grid showing who looks comfortable and who looks stuck. One child's history, for parents' evening. No targets. No time-on-task table. A square in that grid would stay uncoloured until there was enough behind it to mean something, because a report that guesses is a report a teacher learns to ignore.

Nothing in that asks a teacher to type anything in. That is the constraint we would hold ourselves to, and the first thing you should test us on. If a member of your staff has to feed it, we have built the wrong thing.

This is the part that decides whether the rest is worth reading, so it comes before the pitch rather than after it.

"Pupil data held by us: none" is true today for one reason only: there is no server. It is not a policy we are being clever about, it is a consequence of the product being local to the device. We hold no pupil records, no accounts and no work, and nothing is sent to us. Like any website, our web host logs ordinary requests, and those include IP addresses. That is the whole of it.

A class spread across thirty devices, seen from a teacher's laptop, needs a server. On the day that exists we would be holding pupil data, and the sentence above stops being true. We would rather tell you that now than have you find out in week three of a pilot. Any supplier who tells you their classroom product needs no central data is describing something else.

What we would intend — and this is intention, not a fact we have earned yet — is that your school would be the data controller and we would be the processor, acting on your instructions and nothing else. That has to be written down activity by activity before anyone should believe it. The ICO audited twenty-eight edtech providers this year and found most had become controllers without realising, so it is a claim to be checked rather than taken.

We would ask for first name, last name and class. Nothing else. We would not use anything a child does in your school to improve our product.

On profiling, plainly. The games get harder as a child wins, and the next puzzle is chosen from that child's own history. That is an automatic judgement about a child, and your DPIA should treat it as one. What we do not do is show it to the child, rank children against each other, or use it for anything but choosing the next puzzle.

We cannot tell you this makes children cleverer, better at school or more focused. Nobody has measured it, and until somebody has, we will not say it. There is published research behind the design — on what expected rewards do to a child's motivation — and it is set out with citations on our About page. It is research about children in general. It is not evidence about this product, nobody has endorsed us, and no researcher is affiliated with us.

We cannot show you another school using it, because there isn't one. We cannot give you a price, because none has been set. We cannot tell you when the classroom version will be finished.

What we can tell you is a fact about the software rather than a claim about children: nothing in it is built to keep a child there. Whether children come back to games with nothing to earn is the thing we are still finding out.

What we want at this stage is a conversation, not a purchase order. An hour with someone who runs a primary school, about three things: whether the narrow use described above is worth anything in a real week, where it would actually get used, and what would make it useless.

We will not ask you for money, and there is nothing we could send you if you offered. There is no launch date either. The family product is being tested by one family — ours — and it comes first. If that testing tells us children do not come back to games with no rewards in them, there will be no classroom version at all. That would be a fair outcome and we would say so here.

If you would rather look before you talk, the family version is live and you could use it yourself this evening. Payments are switched off while we test, so nothing is charged. When they are switched on it is £6.99 a month for up to four people, or £59.99 for the year, with the first week free. That is the family price. There is no school price, because there is no school product.

Write to hello@pomble.co.uk. Tell us which school you are at and what you would use it for. It goes to the founder — there is no sales team, because there is nothing to sell.