Life GPS Try it on iPhone

Education cannot keep up with the world it is preparing children for.

Right now, governments are restricting what children may use. Other governments are accelerating the same technology into classrooms. Both are reacting to the same fact from opposite directions.

Start with what anyone can verify today.

In short. A child spends their most formative years being prepared for a world that can change faster than educational architecture can. This is our open proposal for another structure: what should organise learning instead, how the real world enters it, where AI belongs, and where the legal and human limits have to sit. It is designed and documented, not deployed — and we do not claim outcomes that do not yet exist.

Three restrictions, all in force at once

Access to AI is being restricted. New York City holds a moratorium on student-facing generative AI from 2K through grade 8 — close to 600,000 students — while high schools keep a limited, vetted use. Norway keeps it largely out of primary school, easing it in from grade 8. France allows no direct use before 4e, then supervised classroom use, then autonomous use at lycée inside a framework the teacher defines.

Phones are leaving the school day. France, England, Brazil, New Zealand and Poland have all moved to national restrictions of different kinds. England's is statutory guidance under the Children's Wellbeing and Schools Act 2026, in force since 29 June and applied by schools from 1 September 2026; it covers the whole school day, breaks and lunch included.

The attention economy itself is being restricted. From 10 December 2025 Australia requires major platforms to take reasonable steps to stop under-16s holding accounts. The stated reasoning includes persuasive design built to keep children online.

And elsewhere, the opposite. China is working toward deep integration of AI across all levels of education by 2030. Japan runs generative-AI pilots down to elementary school. Australia has published a framework for responsible use rather than a ban.

Australia appears in both lists. That contrast makes the distinction visible: restricting systems built to capture attention does not require rejecting tools that can extend capability.

The world agrees there is a problem. It does not agree what the role of AI in childhood should be.

That unresolved space is where this page begins.

This is not about bad schools

The problem is not that the curriculum needs another update.

The architecture updates slowly by design.

An architecture organised around fixed curricula, age cohorts and institutional stages cannot synchronise with a fast-changing technological world in real time. That is a property of its structure, not a failure of the people inside it.

Why it became urgent now

AI now changes tools and methods faster than any curriculum cycle can follow. Two things arrived at the same moment.

Education is short of people. UNESCO projects a shortage of 44 million teachers by 2030.

Technology is reorganising people. One widely used layoffs tracker records more than 200,000 workers affected across the technology sector in 2026, with AI or automation cited in roughly half of the recorded events. It is an industry tracker, not an official statistic. Employment growth in coding-intensive occupations has slowed sharply since generative AI arrived, and the entry level has narrowed most.

This does not mean engineers have become unnecessary. It means the role of technical expertise is being reorganised faster than institutions know what to do with it.

One system is short of people. The other is releasing and reshaping people who hold some of the world's most current technical experience. There is no protocol connecting the two.

These are not interchangeable workforces, and we do not propose replacing teachers with engineers. But a profession being reorganised does not erase the years of judgement, mistakes, craft and contact with reality behind it — and in this architecture that experience does not have to retrain as teaching to be used.

Two of the four roles exist for it. The engineer builds and maintains the systems of the environment. The practitioner carries what a field contains right now — its real problems, its constraints, its entry points. Neither has to become a full-time educator; sometimes a team needs twenty minutes of exactly what one person knows.

The same intelligence that automates parts of work can route human experience to the moment another human being actually needs it.

Meanwhile children have been handed intellectual tools with capabilities schools did not have to plan around even a few years ago, while the systems around them oscillate between prohibition and no supervision at all. The question underneath all of it is narrow:

Can current technical and professional knowledge enter a learning environment continuously — without replacing the teacher, and without waiting for a curriculum cycle to catch up?

Where we draw the line

We agree with the direction of the concern. We do not think the answer is more screen time. And we do not think the long-term answer is keeping children permanently less capable than the tools around them.

Our advantage: Life GPS considers a child's protection in the context of their life — learning, family, health, time, interests and development. These areas affect one another, so protection and education belong in the same picture.

Attention technology and capability technology are not the same thing.

A feed becomes more successful the longer a child stays inside it. AI can be built for the opposite outcome: understand the question, find the knowledge, the person, the laboratory or the next real action — and return the child to the world.

Less phone. More world. Less feed. More creation. More capability, without surrendering judgement.

AI is not a place a child goes into. It shortens the distance between wanting something and reaching the real thing.

Phone bans point the same way we do — away from the screen, towards the world. What they cannot do on their own is give a child something to walk towards. That part is ours to build.

Why this is possible only now

General-purpose AI now makes it technically plausible at consumer scale to take a person's intent apart, find what answers it, connect it, and keep updating as reality responds.

That is not personalised content. It is a continuously assembled environment — and it makes a different architecture possible.

The change is not a faster programme

Instead of updating the programme faster, stop using the programme as the primary way a person's life is organised.

The national programme does not disappear as a public obligation. Required outcomes stay required, and stay verifiable. What changes is that the programme is no longer the main organising unit of a human being's time, attention and development.

The unit of education is no longer the course. It is a human being moving through reality.

What AI actually changes here

This is not an AI tutor personalising a curriculum. That is the old system, fitted more tightly to the child.

AI does not personalise the programme around the learner. It helps assemble reality around the learner's movement through life.

A personalised course still assumes someone already decided what should be learned, in what order, toward what end. Assembling reality assumes the opposite: the next step comes from the person, and the world is brought to meet it.

Its role is infrastructural. It can hold a changing picture of what is happening, notice what a project lacks, take a dream apart into real requirements, search across knowledge and networks, expose constraints, and keep bringing the relevant part of the world within reach.

The more capable AI becomes, the less education needs to be organised around storing and delivering information. The scarce things become different ones: a real desire, human judgement, relationships, access to reality, and the ability to turn an impulse into an actual next step.

If this works, here is what changes

This is what the architecture is built to do. Each one links to the part of this page that shows how.

The unit of education is different. Not the programme, not the class, not the subject, not the institution. A human being moving through reality. Everything else — the timetable, who is in the group, which knowledge arrives, which industry takes part — assembles around what that person is actually doing now, and reassembles when the direction changes. → how it is supposed to work

The walls come down. School, university and work stop being separate stages and become one field. A seven-year-old and a working practitioner can belong to the same world without holding the same role, the same authority or the same responsibility. The same person can be a learner in biology and a carrier of experience in software, on the same day. → how it is supposed to work

AI is neither the teacher nor the personaliser. It is coordination infrastructure between one person and an enormous outside world: it notices what a real attempt is missing, and helps find the knowledge, the person, the laboratory, the permission or the next step. Then the real action happens outside the model again. → how it is supposed to work

The world can find the person. Access stops running in one direction only. An industry does not wait at the exit for a graduate; it keeps a living map of its own reality, and a real opportunity can move toward work that became relevant to it — never toward a searchable profile of a child. → how it is supposed to work

Changing direction does not erase what came before. The mathematics was real, the people were real, the thing made was real, the limits found were real. A child who spends a year on rockets and discovers it was physics, or the sea, or working with a team, carries all of it forward. Changing direction still costs time and effort — but nothing lived is written off. → how it is supposed to work

The environment gets richer with every attempt. What accumulates is the map of reality, not the profile of a child: which laboratory helps with which test, which permission takes three months, which tool everyone recommends and nobody can use. Whoever enters that field next starts from a richer world than the person before them. → how it is supposed to work

The architecture is protected against itself. An impulse is not treated as truth. A company cannot grow itself a recruitment pipeline. A parent does not assign a child a talent. AI does not decide a path. Our own software is never required. A child's data is not the price of taking part. A person with a disability never becomes somebody else's learning project. → how it is supposed to work

That last one is the difference between an architecture and a utopia: the mechanisms against its own capture are part of the design, not a policy added afterwards.

I grew up by doing what I wanted to do and following it as far as it would take me. I learned the world through my own desires — and I learned myself through what happened when those desires met reality.

I think AI makes it possible to build education around that principle for everyone.

— Alexandru Goroh, founder

The order this replaces

For most of history, education worked by bringing the world to a person in advance.

Someone decided what the world contained. They divided it into subjects, ordered those subjects by age, and asked a person to learn them before knowing when — or whether — any of it would matter.

That order existed for a reason. Knowledge was expensive to move, teachers were few, and the only way to prepare a person for an unknown future was to hand them a map drawn by somebody else.

A person can now begin with something they genuinely want to understand, make, change or experience. The environment can then assemble the part of the world that this desire actually touches: knowledge, people, tools, places, money, permissions, institutions, and real consequences.

The person discovers the world by moving through their own desire. And through the world's response, they discover themselves.

Education stops being preparation for life. Life becomes the medium of education.

One field. Everyone who belongs to it.

Take space.

Imagine space exploration as one living environment rather than a subject on a timetable.

A seven-year-old says: I want to build rockets. Somewhere else in that same world, a twenty-four-year-old is working on propulsion. A university has a laboratory. An aerospace company has a real unsolved problem. An engineer knows exactly why a particular design will fail. A working pilot knows what a machine feels like when theory meets the air.

Today these people live in different institutions and mostly never see one another. Here they belong to the same field.

Not the same classroom. Not the same legal role. Not the same level of responsibility. The same living world.

You just want to build rockets and understand space — and it does not matter whether you are seven or twenty-four. It is one environment, gathered by interest. Your colleague may be someone who actually flies. And you can talk to them.

— Alexandru Goroh

The seven-year-old enters at the edge they can actually reach. They build. They ask. They fail. They learn enough mathematics because the rocket demands it. They meet people further along the same reality, and a practitioner may answer a question or open the next door — inside the safeguarding rules of the environment.

At twenty-four the same person may still be there, now contributing to someone else's project. At forty they may be the practitioner answering.

There is no moment when education ends and the real field begins. It was the same field from the start.

And if tomorrow you understand that you no longer want this — choose what you do want. It does not matter where you are from, how old you are, or what you did before.

The environment reassembles around the new direction. Nothing that was lived has to be written off as wasted education.

One picture, in detail

A child says in the morning: I want to build a rocket.

The environment is designed not to answer with a lecture, but to assemble reality around that: to take the dream apart into what is actually needed, then find the people, the knowledge, the tools, the space, the equipment, the money, the experts, the permissions, the organisations — and the next real opportunity after this one.

This is not matching people. A system that only matches people is a professional network for children, and that is not this. Most of what a project needs is not a person.

The chain:

impulse → the dream taken apart → the knowledge, roles and resources it needs → a search across the network → a team → the first real slot of work → making → new gaps → new connections → a result → the next level

How the response is meant to work. A short notice goes out across the network: a team needs someone who can do X — can you help? Whoever answers forms the working slot, and the work starts from there. That is the proposed flow, not a description of something running today.

Reality does not come divided into subjects

A rocket does not know where physics ends and chemistry begins.

A medical device does not separate biology from engineering, regulation, design, communication and money. A problem in a city may need mathematics, politics, materials, psychology and logistics in the same afternoon.

Subjects remain what they are — bodies of knowledge, carefully built. They stop being the primary container of a person's experience.

The real thing being attempted is what reveals which knowledge is missing. That knowledge then enters because it became necessary, and it stays because it was used.

What a parent should be able to hope for

Your child does not have to know at fourteen what they will become at forty.

They do not have to choose the correct track early enough not to fall behind. They do not have to be good at the same things, at the same age, in the same order as everybody else.

They can become fascinated by rockets, spend a year building them, and discover that what actually drew them was physics — or working with a team — or the ocean beyond Earth — and move.

Nothing was wasted. The mathematics they needed was real. The people they met were real. The thing they made was real. The limits they found in themselves were real.

And when the direction changes, the environment changes with them.

The promise is not that a child will choose correctly. The promise is that they are allowed to keep becoming — without getting lost.

One person, over time

Seven. I want to build rockets. Two other children want the same thing. An engineer is nearby. There is a first task for tomorrow.

Nine. The first real model does not fly. Mathematics arrives, and so do materials, because the failure asked for them.

Eleven. The team wants to test an engine. A laboratory appears, and with it the safety rules that come with real equipment.

Thirteen. Engines stop being the interesting part. What holds them now is how anything stays alive in an extreme environment. The path turns toward biology, and the environment turns with it.

Sixteen. A small piece of real research, alongside a university group.

Nineteen. Out into a laboratory in the world, paid for part of the work.

Twenty-two. A research career turns out not to be it. Back into the environment, and onward into design — or the ocean.

Thirty-five. A notice arrives: a team needs someone who understands X — can you help? This time the person answering is them.

There was never a moment when this person finished their education. There was never a moment when their life had not yet begun.

Not a new school. A new organism.

This is not a school with a different timetable. It is a living organism built from things that already exist but sit today in separate systems.

A school has rooms and adults. An industry has current knowledge and real problems. A university has laboratories. A municipality has infrastructure. A foundation has capital. A company has tools. A community has people. Government defines required outcomes and safety. AI can see the gaps between them.

Those pieces almost never assemble around one person's actual next step. The reform is the protocol by which they do.

The child does not travel from institution to institution trying to fit each one. The institutions, people and resources temporarily assemble around what this person is actually trying to do — and then re-form around the next thing.

The world can find the person too

Until now, access has run mostly in one direction: the learner searches for the institution, applies, proves eligibility, and asks to be admitted.

A living field can work both ways.

An agency, a laboratory, a university or a company may publish a real problem, validate the map of a field, or offer a supervised opportunity. A project inside the environment may reach the point where that opportunity becomes relevant. Then the door can move toward the person.

Not because an algorithm labelled a child "gifted". Not because an organisation bought access to talent. Not because children became searchable profiles.

Because something the person actually did became relevant to something real in the world.

Sharing identity and contact details stays limited by age, child-protection rules and whatever permission the law requires. The work may become visible where the human being must not.

The open standard is designed to expose an opportunity — and, where a lawful basis exists and any guardian authorisation required has been obtained, relevant work. It may not expose a searchable profile of a child.

There is no graduation from becoming

In a living field, learner and contributor are not permanent identities.

A seven-year-old can know something useful that another child does not. A twenty-four-year-old engineer can be a practitioner in software and a complete beginner in biology. A working scientist can teach for an hour and spend the next hour learning from someone else.

Competence changes by domain and by task.

A person does not graduate out of learning and into usefulness. Both run the whole way through a life. What protects a child inside that is not a lower status — it is the legal and child-protection requirements described further down.

One core. Three physical forms.

Start with an existing school. Existing teachers, existing rooms, the phones people already have, the local community. The reform adds protocols and human coordination; a digital mediator is optional. No new building, no new category of staff and no laboratory is a condition of entry.

A partially transformed school. Project spaces, changing teacher roles, local practitioners and a wider network, where resources allow.

The full learning environment. Laboratories, making spaces, social life, and residential infrastructure where it is wanted.

If it cannot work in an existing school, the model has failed its own test. A reform that needs the richest version before it can begin is not a reform for most of the world.

The world becomes the campus

The long-term aim is not to build a special campus everywhere. The full campus is one physical form of this, not the point of it.

A school room. A workshop. A farm. A university laboratory. A museum. A factory. A local engineer. A municipality. A research group on another continent.

They already exist. What is missing is the ability to assemble the right part of that world around a person at the moment it becomes relevant.

AI makes that coordination increasingly possible. Human beings remain the reason for doing it.

The campus, where it goes that far

A place to live in, not an institution to attend. Sleep, food, showers, rest. Something between a good office, a summer camp, a workshop, an incubator and a home.

Better to disappear there, making something, than in the places people disappear into otherwise.

Social life is part of it rather than a distraction from it: friendship, first attraction, the ordinary business of growing up among people — the things that made summer camps matter, put back together with real work.

Spaces are separated, not ages. In project and social space, mixed ages are the point. In personal and living space, different rules apply and age boundaries are explicit.

What this requires, and what is written into the design: care and hygiene standards, supervision ratios, facilities and licensing, insurance, and the health and safety law of each jurisdiction — defined before a pilot opens.

How this does not become chaos

The rest of this page is the part that matters more than the vision: what keeps a human being in control of it.

The impulse starts the process. It does not rule it.

An impulse is taken seriously, not obeyed blindly. A child can copy a friend, absorb a parent's ambition, follow status, be shaped by an algorithm, or simply want something dangerous.

The environment's job is to turn an impulse into a safe experiment against reality: try it, meet what actually happens, see what answers back — then go deeper, or leave it without shame.

Parents set the legal and safety boundaries. Curators protect the person. Practitioners bring reality. None of them — and no company, funder, government or AI — gets to assign what a child should want. They may open possibilities. They may not manufacture the impulse.

AI is not the teacher

AI is not the authority at the centre of this environment.

It does not decide what a person should become. It does not manufacture an interest, optimise a child toward a labour market, or replace the adult who actually knows the child.

What a child may hand to AI

There is not one use of AI. There are four, and they are not equivalent:

Explain what I do not understand · extend what I can already do · check what I made · delegate the whole thing.

The first three can extend capability without replacing the learner's judgement. Delegation is different: it can bypass the very capability being formed — and it is what most of the current argument about AI in schools is actually about.

Where a person cannot yet judge the result, the system does not produce the result for them. Where they can, it works as an amplifier.

Role and depth of participation follow demonstrated capability and real contribution. Age, law and safeguarding remain hard external boundaries that no demonstrated capability overrides.

Four roles, not two

Answers for
The person the impulse — everything starts here
The curator the human being: relationships, development, conflicts, culture, boundaries
The engineer systems, technology, the currency of the world, tools, automation, integrating whatever appeared this month
The practitioner what is actually needed in this field, from someone who works in it

"Curator" is a functional role in this architecture, not a new legal profession. In each jurisdiction its responsibilities must be mapped onto an existing, legally recognised and qualified role inside the institution — or created only where the law permits. Calling someone a curator gives them no authority the law does not already give them.

The engineer does not decide what a cosmonaut needs — the field of cosmonautics decides that. The engineer turns the knowledge of an industry into a living, working system of development.

The design redistributes teachers rather than replacing them. We deliberately reserve the human responsibility — development, relationships, culture, judgement and boundaries — for people, rather than delegating it to software. That is a choice we are making, not a prediction about what machines will never manage.

This is also why the engine does not go out of date: a new instrument appears, and within weeks it is inside the directions that need it, put there by people who use it for a living.

And the parents

A child is not only shaped by what they choose. They are shaped by what the people around them expect, fear, praise and quietly rule out.

Part of this design is a protocol for parents — support for the adults closest to a child in helping them unfold rather than narrow: how to widen what is possible instead of closing it early, how to tell encouragement from pressure, how to let a direction change without treating it as failure.

The parent's role is observation before interpretation. The design deliberately separates what a parent actually sees from the story a parent may already hold about who their child is, or should become. Where the optional family layer of Life GPS is used, that becomes structured observation and questions — never a parent assigning a talent, a profession or an identity to the child.

Designed, not yet built. It belongs to the same review as everything else involving children.

Everyone is included

Non-speaking children. Children with special needs. Any form of participation.

The design may not exclude a participant because they need additional support. Local curation is arranged for whoever needs it, on the spot.

A participant's own need can become a project of the environment: a child needs a prosthesis, and making it becomes real work for others — or it is bought from a dedicated fund for adaptations and special needs. Both routes exist on purpose.

The person with the need is never the material of somebody else's education. If an adaptation becomes a project, that person stays the client and the co-author, may refuse the project entirely, and their dignity outranks the learning opportunity.

How learning is organised

No grades. There are levels — what a person can actually do on their own inside a real situation. Not the volume of what they know, not their age, not hours logged. In rockets one level, in music another. Nobody is ranked against anybody.

Age is not a principle of distribution. People come together by impulse, interest, direction, task, and what they can contribute. Everyday, legal and protective boundaries do take age into account.

Took it on — close the minimum. Not as punishment: so a person can tell the difference between I have stopped wanting this and this became hard and I am running away.

Required national competencies are not cancelled. They are laid out inside what a person is already doing, wherever the work genuinely covers them. The result stays mandatory; the path is individual. Whatever a project does not cover remains an explicit obligation — it does not quietly disappear into "personalisation".

Groups and curators re-form constantly around live tasks, instead of standing fixed for a year.

No wall between learning, working and living

School, university and work are not separate stages here. It is one continuous environment.

A person can go out into the world, work, earn — and come back. That is a normal state, not a failure. Not a student, not a founder, not unemployed: a person doing what pulls them, who is never left alone.

The school gate is not the edge of the project

A real project eventually meets the real world: permissions, laboratories, municipalities, universities, companies, testing, certification, customers and law.

The environment does not simulate that boundary. It helps the team find out what is actually required to test, show, register, build, sell or deploy what they made.

For minors, commercial activity, ownership and compensation remain subject to child-rights, labour and intellectual-property law. Those rules must be met; a project cannot bypass them.

Industry validates the map of a field, not the worth of a child

Practitioners help describe what a field actually contains now — its branches, tools, constraints, entry points and real work — and challenge that model before it ever enters a pilot. Afterwards they keep it current, which is how industries end up growing their own next generations.

A company may contribute equipment, expertise or a real problem. It may not shape children's interests toward its hiring needs, buy influence over a direction, rank participants, or turn the environment into a catalogue of talent.

Two layers: the person, and the world

The educational environment works outside the person: people, knowledge, tools, laboratories, schools, companies, universities, money, permissions, laws, opportunities, places.

Life GPS, where a person or family chooses to connect it, is the other layer — the continuity of the person across the whole of life. What matters to them. What they are attempting now. What changed. Their relationships, their constraints, their earlier choices, the way they want to be helped, and the parts of their life that do not disappear the moment they walk into a school.

The two layers must never collapse into one another.

The educational environment does not own the person. Life GPS does not own the educational environment.

The reform assembles the outside world around a real next step. Life GPS can carry the continuity of the person into that assembly, and carry what happened back out into the rest of their life. Where it is not used, the reform still works — through people and the open protocol.

Where age, law and the required authorisation permit, a child or a family can connect it directly. Then the child has a continuity layer that belongs to their life rather than to their school: change school, city, project or direction, and the personal context does not reset along with the institution.

A school knows a child at school. A field knows them inside a project. A parent knows them at home. A practitioner knows them in one particular craft.

Institutions own their responsibilities. The person owns their continuity.

That is how this organism can be enormous without becoming one enormous system of observation: no institution in it ever holds the whole model of a human being.

A living environment has to react

A living system cannot wait until the end of a term to discover that reality changed three months ago.

Something always changes first. A new technology appears. A person's interest moves. A project gets stuck. A team is missing somebody. A child and a parent start pulling in different directions. Friction begins to build. Someone becomes overloaded. A real opportunity opens. A project hits a legal boundary. A participant needs an adaptation the environment does not have.

Each of those is a trigger. A trigger is not a diagnosis, a score or an instruction. It is a reason for the environment to ask one question: what changed — and what should now be assembled differently?

What changed What the reform does What Life GPS may add, if connected
A technology appeared practitioners update the living map of the field brings the new possibility into a person's path when it becomes relevant
The person wants something else team, resources and route reassemble around the new direction keeps continuity, so changing direction is not a fresh start from zero
The project is missing something finds people, knowledge, tools, money, space or permissions can surface a connection the person could not have seen
The same difficulty keeps returning the curator helps the person distinguish the friction of learning from a direction that is genuinely over holds the wider life context before another demand is added
A tool can do more than the person can yet judge limits on what may be delegated, and a named human who reviews the result may expose the constraint; it never decides alone
Parent and child diverge the child's own voice, and the parent's observation protocol, stay separate the mediator may surface what matters to both sides before it hardens
Friction starts inside a team human mediation by the curator, before anything escalates each side can see the other's context
An impulse turns unsafe boundaries, law, safeguarding, and a safer version of the same experiment may expose the constraint; it never decides alone
A real opportunity appears the open standard brings the opportunity toward the relevant work limited by law and permission; no catalogue of children
A special need appears local curation, the adaptation fund, the environment changes carries needs and preferences across settings, if the family opts in
The project reaches the real world laws, licences, testing, institutions and customers enter helps hold the next lawful step inside the person's wider path

Before friction becomes an incident

Safeguarding often has to respond once a concern has already become visible.

We want to test whether some signals can be surfaced earlier — without turning a child into a monitored score.

A disagreement that keeps coming back. A child withdrawing from their team. A parent and a child describing the same situation in opposite ways. Repeated overload. A pattern of boundaries being crossed.

Where a family has chosen to connect it, Life GPS could offer an additional voluntary route to say that something is wrong, and the mediator could bring the relevant sides into the same context before friction becomes escalation.

But a trigger is never a verdict. It does not diagnose the child. It does not accuse an adult. It does not punish anything by itself. It starts a protocol: check, listen, involve the right human being, keep the voices separate, and escalate only where the safeguarding rules require it.

Life GPS may be an additional route to protection. It can never be the only one. A child without our product must have exactly the same right to be heard and protected.

And to be explicit about what it is not: Life GPS is not a medical, emergency, crisis or safeguarding service. It does not diagnose, does not treat, and does not replace the people and institutions that carry legal responsibility for a child. Where there is danger, its job is to get a human being involved faster — never to handle it itself.

Safeguarding comes before pedagogy

An environment where adults from outside work alongside children answers this first.

This safeguarding architecture has not yet passed independent child-protection review. The rules below are design constraints for any pilot — not claims about a deployed system. They are published now so that people who know more than we do can argue with them before anything involves an actual child.

With that said, the constraints themselves are not soft:

Children's data

Design constraint: children's data may not become the business model, and may not be sold or traded. Any disclosure required for safeguarding or by law must be purpose-limited, minimised, and defined before deployment rather than decided in the moment.

The pilot is designed to keep personal data primarily on the person's device. Legal grounds, child-protection rules and limits on data collection must be defined for each country before a pilot starts, then checked independently.

This is a written constraint, not a delivered or audited one — there is no deployment yet in which it could have been tested. We state it now, in these words, so that it can be held against us later rather than softened.

Every project should leave the environment better for the next person

A project leaves behind more than the thing it made.

It reveals that the map of a field was missing a branch. That a particular laboratory can help with a particular test. That one permission takes three months. That a tool everyone recommended does not actually work. That a practitioner is willing to help again. That a safety rule was incomplete.

Those lessons improve the environment itself.

The person learns from the world. The environment learns from the person's attempt. Whoever enters that field next starts from a richer reality than the one before them.

Personal data does not need to be part of that commons. What compounds is the map of reality, not the profile of a child.

The fund

A bridge from an idea to a physical thing: micro-grants, equipment, materials, access to laboratories, experts. Without that layer children only make what can be made from cardboard and a laptop.

Funding follows a viable next step of a project — not marks, not performance.

A separate fund covers adaptations and special needs, so that "everyone is included" is a budget line rather than a sentiment.

What one cycle leaves behind

No single score of a human being.

The learner has work they actually made, levels in specific domains, the teams they worked with, and a clearer signal about where they want to go next.

The school can see which directions formed, which teams held together, which artefacts reached the outside world, and which required competencies were genuinely covered through that work.

The public system can verify required educational outcomes without receiving a profile or a ranking of any child.

The field itself can see whether its model of the domain was wrong — and the model is revised.

The coordination layer is replaceable by design

The environment has two ways to assemble what a project needs.

Human assembly — curators, practitioners, a visible board of missing resources, team fairs, people finding each other directly.

Digital mediation — software able to see that a team needs X, and that the wider network holds people or resources that might answer, and to propose the connection. The people decide.

Underneath that distinction are three layers: the open reform, a proposed open interface for any implementation to use, and Life GPS as one optional implementation of mediation, memory and coordination.

If a function of the reform can only be performed by our software, we consider that a defect in the reform.

Joining the environment never requires installing Life GPS. A school may run the reform without our software. An institution may use another implementation of the open interface. Any family-facing digital layer stands on its own lawful basis and its own authorisation, separate from the school's. Participation in education and participation in our product are two different decisions.

Open core

Countries, faiths, schools and companies fork the core into their own distributions — their language, their law, their curriculum. We maintain the core and hold the name.

A fork may change the distribution. It may not quietly change the constitutional core and still present itself as the canonical system.

Publishing code is not the same as making something last. Governance, compatibility between forks and long-term maintenance are part of the design problem; an abandoned fork helps nobody.

And the rule we hold ourselves to everywhere applies here with the most force: the environment may open possibilities to a child. It may never shape a child's impulse toward us.

Precedents exist for the components

None of the pieces is ours, and we say so:

Mixed ages — Montessori, since the first Casa dei Bambini opened in January 1907; 119 years. Learning through interest — Big Picture Learning: 275+ schools, fellows and places of learning, interest-led internships, a long-term advisor relationship. An environment without teachers — the 42 network: 50+ campuses in 30+ countries, no classes, no professors, peer to peer. It is adult software education, and it is not evidence for a model involving children. No grades, no fixed curriculum — Sudbury schools, for decades.

The proposal is the architecture that connects them, with the walls between them removed. We have not demonstrated that the combination works. That is what a first pilot is for.

What this is not

We are not saying school should be abolished. The reform begins inside existing schools, with existing staff and existing obligations.

We are not saying the national curriculum is cancelled. Required outcomes stay required and stay verifiable.

We are not saying AI will teach children. AI is infrastructure here; the adult who knows the child is not replaceable.

We are not saying a child always knows what they want. The impulse starts the process and does not rule it.

We are not saying learning should only be pleasant. Took it on — close the minimum.

What we are saying is more radical than any of those: the unit of education is no longer the course. It is a human being moving through reality.

What is deliberately not solved on this page

We have not fixed the recognition model that would let the outside world read domain levels without recreating the diploma system.

We have not fixed ownership and compensation when a minor creates something of commercial value.

We have not fixed admissions and funding — who can enter, and how this avoids becoming an environment only wealthy families reach.

We have not closed the child-monitoring and intervention architecture. That requires independent child-protection, legal and developmental review.

We have not fixed the independent advocacy and escalation route for a child when the problem is the curator, the institution, a parent — or us. A safeguarding system cannot require a child to complain through the authority they may need protection from.

These requirements must be met before launch.

Where this actually stands

Designed and documented. Not deployed. No pilot school. No campus. No learner outcomes claimed, because none exist.

Life GPS exists today on iOS, and first users are testing it before launch — see and install the working product. That is evidence we can build something of this size. It is not evidence that this educational architecture works.

What is assembled: the core principles; the participants and what each may and may not do; the lifecycle of one school from first cluster to full environment; the research map, with what has been verified and what was struck out; and the requirements that must be met, in order.

What is not frozen: the exact boundary between the open core and a national distribution — which principles no fork may change, and which parts belong to law, language, curriculum, culture and faith in each jurisdiction. That has to be explicit before the core is published.

What remains: independent child-protection and legal review; closure of the data and trigger-response architecture; the deployment package; implementation; an independent evaluation design; validation of the first field models with practitioners; and the first bounded pilot at one or two institutions.

It starts after the first round of the main product — or earlier, if the reform raises funding for its own first stage. That is what we are looking for now, and what it buys is set out below. If you want to talk about it: hello@lifegpsos.com · @LIFEGPS_SUPPORT_BOT.

The reform is open and free to adopt, and it is designed to work without our software — a constraint of the design, not a concession. That does not mean physical implementation costs nothing: projects, adaptations, safeguarding, evaluation and richer environments all require resources.

Working name LIFE GROWUP; the final name is not fixed.

How it starts — the smallest lawful pilot

Not from a ministry, and not from a new building.

One willing school. The smallest number of clusters the jurisdiction permits. Existing buildings, existing staff, existing curriculum obligations. The lawful basis, parental or guardian authorisation where required, and every institutional approval the jurisdiction demands. No new campus is required: the entry model is designed around buildings, staff and obligations that already exist.

Legal form is jurisdiction-specific. Czechia is our first legal mapping; every pilot gets its own education-law review before launch.

Entry has to be reversible. A participant who wants the ordinary class back returns without being treated as a failure. A school that decides the method is not working stops it, without being trapped by infrastructure, software or contracts.

The first pilot begins in the smallest legally permitted form, and expands only after one cycle produces evidence worth expanding.

And the deployment pack has to exist before any of it: role protocols, document templates per country and its law, the lawful-basis, notice and parent or guardian authorisation documents that jurisdiction requires, deployment instructions, how to monitor, how teams are formed.

What the first funding actually unlocks

We are not raising money to build a global network of schools.

The first funding addresses the requirements that must be met before one lawful pilot can open:

independent child-protection review · education-law and data architecture for the pilot jurisdiction · the trigger-and-response protocol — which situations the environment must recognise, what it is permitted to do about each, who the responsible human is, where escalation begins, what data is involved, who sees it, and what happens when there is no Life GPS at all — independently reviewed before any pilot involving children · the deployment pack for one school · an independent evaluation design, with the conditions for failure defined in advance · validation of the first field models with practitioners who work in them · the initial project and adaptation funds · and running the first bounded pilot.

The leverage is the package, not the building. If the pilot works, what becomes reusable is the protocol, the evidence, the legal and deployment knowledge, and the open core — not a requirement to finance another campus from zero.

What we are looking for

A school leader who wants the smallest lawful pilot inside their own school.

An independent research or evaluation partner to define before the pilot what should be measured — including what would count as this model failing.

An implementation partner who already knows how to work with public systems and ministries.

A funding or philanthropic partner for the first stage only: design, safeguarding, evaluation, one pilot. Not for a global reform.

Contact: hello@lifegpsos.com · @LIFEGPS_SUPPORT_BOT

In the end

A child says they want to build rockets. Somewhere in the same field, a person who actually flies is at work, a laboratory has equipment, a company has an unsolved problem, and an engineer knows why a design will fail.

None of that has to stay on separate sides of an institutional wall anymore.

For as long as anyone can remember, we asked children to spend their best years learning a map of the world drawn by somebody else, in an order chosen by somebody else, to be used at a time nobody could name. It was the best we could do. Moving knowledge was expensive, and there were never enough teachers.

That constraint is lifting. Not because a machine will teach children — because coordination has become cheap enough that the world can be assembled around a person instead of a person being sorted into the world.

What that buys is not efficiency. It is permission.

Permission for a child to want something badly, follow it as far as it goes, meet reality, and be changed by what they meet. Permission to stop, and to choose again, without the years behind them being written off. Permission to be useful before they are finished, and to keep learning long after anyone would have called them finished.

That is what the whole design protects: a person's right to keep becoming — and to never be lost while they do it.

Nothing here is proven yet. One school, one cycle, and an evaluator who was allowed to define failure in advance will tell us more than any argument on this page.

But if it does work, the change is not that education gets better. It is that education stops being the thing you do before your life, and becomes the way a life is lived.

Sources

Generative AI and children: restriction and acceleration

Phones, feeds and accounts

Precedents

Life GPS · lifegpsos.com