Карта влияний на ИИ
Какие силы определяют развитие ИИ — и что из этого необходимо учитывать, создавая правила его действий?
Пройдите по вкладкам: основные силы, участники и пересечения — не полный перечень мировых влияний. Дополните то, что мы упустили.
From understanding the world to working together. The map above shows influences and dependencies we need to consider. The examples below explore how existing organisations and services could interact with the person's side — what each could contribute, what information it needs and what we could test together.
What each side exchanges
The human side does not replace institutions or jurisdictions. It lets the person interact with each of them directly, within that institution's legitimate role. No institution has to speak for the whole person to another institution.
Examples of systems around the person
Insurance Earlier human action, not deeper insurer surveillance.
AI is already changing insurance, and that change will continue. The question is whether insurance's architecture becomes part of a shared human-side protocol we can test together.
AI will give insurers ever more signals about people. There is another path. The person sees the road to a claim earlier — sleep, workload, money, deferred treatment — and acts before it becomes your case: uses a benefit already paid for, sees the doctor, asks for a schedule change, reaches prevention or rehab.
What you never get: a whole-person risk profile, hidden profiling, risk or price assessed on your behalf, behaviour tracking tied to premiums, control over the person's route. Your regulated decision stays yours.
What you may get — a pilot question: fewer preventable claims, earlier intervention, better adherence, fewer disputes, more timely use of what you already fund.
How it works: the person's own side holds the trajectory; you receive only the action a specific decision requires. Life GPS is designed to keep that missing context on the person's side and expose only the action your system needs.
What we test together: a corpus of authorised, compliant, locally correct — detrimental to the person insurance cases; then one product line with preregistered metrics. Status: research object, not a live integration.
Employers The signal appears first on the worker's side. The employer receives only the action it needs, not the reason from the rest of the worker's life.
AI is already changing workplace monitoring, and that change will continue. The question is whether employers' architecture becomes part of a shared human-side protocol we can test together.
AI promises earlier warning at work — usually by collecting more about the worker, exactly where labour, privacy and discrimination risk rises. There is another path. The worker sees the conflict across workload, health, family and money first, acts earlier, and sends you only the action required: a schedule adjustment, an occupational-health visit, a leave or accommodation request.
What you never get: whole-person context; health, financial or family feeds; fitness, promotion or discipline scoring; hidden productivity management; the reason behind a request unless the worker lawfully chooses to share it. Clinical and occupational decisions stay with the professional. Workers and unions can verify that support did not become surveillance.
What you may get — a pilot question: a shorter path from fatigue to action, more use of what you already fund — occupational health, EAP, leave, benefits, rehabilitation — fewer incidents and disputes.
How it works: a worker-owned route; you receive requests, not causes. Life GPS is designed to keep that missing context on the worker's side and expose only the action your system needs.
What we test together: an Early Action pilot in one high-risk population against the ordinary support pathway; stop condition — any gain that needs more employer visibility. Status: research object.
Banking The lender does not need the reasons from the rest of a person's life. The person needs those reasons before the lender needs to intervene.
AI is already changing lending through alternative-data scoring, and that change will continue. The question is whether banking's architecture becomes part of a shared human-side protocol we can test together.
AI lending moves toward more data about the customer. There is another path. More context stays on the person's side, so the person acts earlier — postpones a purchase, uses insurance or a benefit, builds a reserve, contacts you before a missed payment. You receive the ordinary lawful request: restructure, flexibility, product action.
What you never get: an alternative whole-life credit score; creditworthiness, eligibility or price decided for you; a hidden vulnerability classification; health, family or relationship context in underwriting. Obligations stay real. Your regulated decision stays yours.
What you may get — a pilot question: earlier contact before arrears, more sustainable resolutions, fewer defaults reached too late, less private data to hold.
How it works: a performing account becomes an ordinary request — not the life behind it. Life GPS is designed to keep the reasons on the person's side and pass you only the request.
What we test together: Earlier Contact Before Missed Payment against ordinary hardship entry; plus a scenario corpus for agent-applied credit that is valid yet wrong for the life. Status: research object; agent-credit path — direction.
Healthcare Doctor stays doctor. The person carries the medical decision into the rest of life.
AI triage and navigation are already changing healthcare, and that change will continue. The question is whether healthcare's architecture becomes part of a shared human-side protocol we can test together.
Patients already arrive with AI answers, and platforms already connect whole-health services. The open question is what happens after the consultation — whether a clinical decision survives the work schedule, the money, the caregiving and the transport that the clinician neither owns nor should manage. Life GPS is designed to carry the decision back into the route: prepare questions, preserve the decision, route to support that already exists, re-escalate when interpretation is needed.
What you never get: diagnosis, prescription or treatment decided by software; a whole-person health profile handed to employers or insurers; clinical authority claimed by a system; a legally required medical fact hidden.
What you may get — a pilot question: better follow-through, fewer dropped recommendations, earlier return to the right professional, less disclosure outside healthcare.
How it works: the clinician receives what is clinically relevant; the employer receives an accommodation request, not a diagnosis.
What we test together: Post-Consultation Follow-Through on one bounded clinical pathway; stop condition — the system interpreting an ambiguous clinical instruction. Status: research object.
Government Government can integrate services around a life event. It should not become the owner of the whole life behind that event.
AI is already changing government services, and that change will continue. The question is whether your rules and infrastructure become a compatible, interoperable part of a shared human-side protocol we can test together — not one more closed ecosystem.
Governments already organise services around life events, and wallets let a person disclose exactly what was consented. Once-only inside government is still not once-for-life across employer, bank, doctor, family and a second country. Life GPS is designed to continue the route where the public service ends: one life event, each agency receiving only what its decision requires, the private consequences staying with the person.
What you never get: social scoring, a hidden citizen profile, any single institution holding the whole person. We do not decide eligibility, tax, immigration or enforcement, and we do not replace identity — we connect with existing public digital infrastructure, without replacing it.
What you may get — a pilot question: fewer repeated forms, faster time-to-benefit, fewer abandoned applications, measurable life-event outcomes where measurement often lags deployment.
How it works: the state receives application, credential, consent; the rest of the event stays on the person's side.
What we test together: one life event inside one existing reform, independently evaluated, locally implemented. Status: research object; wallet-level integration — direction.
Foundations Fund a bounded public-interest object, not preferential access to a product.
AI is already changing public-interest funding and evidence questions, and that change will continue. The question is whether that standard becomes part of a shared human-side protocol we can test together.
The AI transition needs public infrastructure that should belong to no single product: a failure corpus, open protocol work, independent evaluation, affected-person participation. A foundation can fund one falsifiable question — for example, whether earlier person-owned action reaches disability support without increasing employer surveillance — while an independent evaluator keeps the right to publish a null result.
What you never get: placement in the person's route, user access, privileged data, control over conclusions or protocol governance, preferential commercial access. Sponsor ≠ evidence owner.
What you may get — a pilot question: shared evidence and standards that stay useful even if the company changes direction. A null result is still a public-interest outcome.
How it works: an agreed purpose, responsible organisation, expected results, public/commercial boundaries, publication rights and end date — agreed before money. Life GPS is designed to keep the funded object independent of the commercial product it studies.
What we test together: one bounded public-interest object; failure condition — funding that buys influence over route, data or conclusions. Status: research object; some routes wait on eligibility decisions on our side.
Women & family The woman keeps the whole route. Every institution keeps its proper role.
AI is already changing the systems around women and families — health, care, work, finance — and that change will continue. The question is whether this rebuild becomes part of a shared human-side protocol we can test together.
Care, health, work and finance are increasingly funded and governed as connected systems. A woman is at once worker, parent, borrower, patient, partner and caregiver; each institution can act correctly in its own role while the conflict between those roles stays hers to carry. Life GPS is designed to keep that route on her side — work, health, finance, childcare, eldercare, insurance and family each receive only what their own action requires.
What you never get: reproductive plans, fertility, pregnancy or family circumstances passed to an employer beyond lawful specific disclosure; a women's-life underwriting profile; family context turned into family access; an AI that infers reproductive or relationship state and acts on it silently.
What you may get — a pilot question: care, work and health conflicts surfaced earlier, better use of support already funded, less coordination burden, less unnecessary disclosure.
How it works: one route, many minimal handoffs.
What we test together: a scenario corpus across care + work + health + finance, then one care transition; stop condition — coordination that requires a shared family or reproductive profile. Status: research object.
Children Protection is not total surveillance.
AI companions and age assurance are already changing child-safety systems, and that change will continue. The question is whether this sector's architecture becomes part of a shared human-side protocol we can test together.
Child-centred AI safety, age assurance and helplines are active fields, and children already use AI for personal questions. "Protect the child" can collapse into total parental or platform surveillance; "respect privacy" can fail safeguarding. Life GPS is designed to hold the route on the child's side with age- and capacity-sensitive participation, explicit guardian scope and explicit safeguarding escalation — helping the child tell AI suggestion from professional help, prepare a conversation, choose what to share.
What you never get: a whole-child profile; automatic guardian access to every private thought; practitioner or vendor access to the child; interests or distress used as targeting signals; age or safety data repurposed; a claim that software alone determines best interests.
What you may get — a pilot question: clearer escalation, earlier professional or helpline use, better child understanding and control, less unnecessary disclosure.
How it works: predefined, proportionate escalation rules; the child's voice grows with age.
What we test together: a best-interests / escalation scenario set co-designed with child representatives and safeguarding professionals — not deployment to minors first; stop condition — safety that works only through permanent whole-route visibility. Status: scenarios and evaluation now; active protection — a future direction.
Ageing More support without taking away the person's authority.
AI-assisted monitoring is already changing long-term care, and that change will continue. The question is whether this sector's architecture becomes part of a shared human-side protocol we can test together.
More digital help for an older person can quietly become more authority for a daughter, a caregiver, a bank or a care system. Life GPS is designed to make delegation explicit, limited, visible and time-bound: who may help with what, what still needs the person's confirmation, what changes when capacity changes, which earlier instruction remains authoritative.
What you never get: age as a general loss-of-agency flag; a longevity or frailty score; automatic family access; permanent authority because someone once helped; health or cognition inferences reused for unrelated banking, insurance or service decisions.
What you may get — a pilot question: earlier support activation, fewer unnecessary transfers of control, safer high-impact confirmations, better continuity across care, health and finance.
How it works: the caregiver reports an observation, the clinician decides medically, the family acts within a delegated role, the bank follows a rule — no helper owns the rest.
What we test together: the Delegation & Continuity Boundary Set — changing capacity, caregiver escalation, unusual transactions, earlier versus current wishes; stop condition — age or cognition becoming blanket permission. Status: research object.
Disability Nobody should have to reconstruct themselves for every system that was not designed to include them.
AI is already changing accommodation and accessibility processes, and that change will continue. The question is whether this sector's architecture becomes part of a shared human-side protocol we can test together.
A person with a fluctuating disability proves the same functional need again and again — to employer, government, transport, education, insurer, technology provider — usually by resending a diagnosis. Life GPS is designed to keep functional needs, permissions and accommodations on the person's side; each system would receive the concrete action it must provide — large text · remote attendance · avoid Y · accommodation Z — never a reusable disability profile.
What you never get: a disability, health or productivity profile for employers; eligibility taken from government; functional-support data as an underwriting profile; support turned into transferred authority; an accommodation silently reused for unrelated scoring.
What you may get — a pilot question: lower repeated-proof burden, faster accommodation, better retention and access, less diagnosis disclosure across systems.
How it works: functional need and diagnosis are different objects; only the first travels.
What we test together: an Accommodation Portability Test across three or four systems — only after co-design with organisations of persons with disabilities; stop condition — any central disability profile. Status: direction; co-design first.
AI agents Who holds the person's side when several authorised agents are all locally right, and their combined actions conflict with one life?
Agent identity and assurance are themselves changing fast, and that change will continue. The question is whether the standard they converge on becomes one shared, interoperable human-side protocol — or a set of closed, competing ones.
Standards are giving agents identity, delegated authority and tools; assurance is becoming a market. Across today's standards, many current layers answer may the agent do this? — rarely should this still happen for this person? Life GPS is designed to hold a person-side authority that turns private context into a bounded mandate — the agent would receive salary ≥ X · relocation = no · commute ≤ Y, never why.
What you never get: a whole-life model, raw vulnerabilities, hidden reasons, a relationship graph, permanent authority from a one-off consent, the person optimised for a platform. Only allow · deny · ask human · constraint · expiry · receipt.
What you may get — a pilot question: a test class your evaluations do not run — locally correct, globally wrong — and disclosure that is measured, not assumed.
How it works: the design is for constraints and receipts to flow to executors while reasons stay on the person's side; the sequence of requests must not rebuild the profile.
What we test together: Constraint Verification — 20–30 falsifiable does/does-not claims against code — then scenario tests across model, cloud and agent swaps. Status: architecture under test; not claimed live.
Education Learning should not end when the course does.
AI is already changing education, and that change will continue. The question is whether education's architecture becomes part of a shared human-side protocol we can test together.
AI tutors optimise assignments and courses. Too often, after the course the route disappears and what was learned risks becoming only a grade. Life GPS is a continuity layer: a real impulse becomes a real project, the required curriculum is mapped into it, the right practitioner joins without replacing the teacher, evidence of capability is produced — and the route continues into the person's life.
What you never get: curriculum, grades or admissions decided by software; a whole-child profile; teacher judgement replaced; a practitioner with autonomous access to a minor; a child's interest used as a targeting signal. The construction must work without our product — otherwise it is not reform.
What you may get — a pilot question: visible curriculum coverage, evidence of capability instead of scores, less teacher workload, less child data disclosed.
How it works: the teacher stays the pedagogical and safeguarding side; the practitioner answers only for real-world relevance.
What we test together: one school, one lawful project cycle, existing teachers; failure conditions written first. Status: research; outreach on hold.
Pensions A pension system can help a person prepare for retirement. It should not need to become the operating system for the rest of that person's life.
AI is already changing pensions and work-ability systems, and that change will continue. The question is whether pensions' architecture becomes part of a shared human-side protocol we can test together.
A pension system sees contributions, balance, projection, payout. The outcome is formed for decades outside the account — health, caregiving, children, housing, debt, disability. Life GPS shows the person the divergence between work-ability and the life they chose before it becomes a disability-pension case; the next step is a service that already exists.
What you never get: longevity or health inference, family situation, a "good or bad saver" judgement, the right to rewrite the person's goals for the pension pot, worker data for disability-risk prediction. And no nudge engine: the layer must be able to surface that a lower contribution may better fit the person's stated priorities when a child, health or housing matters more — and let the person decide.
What you may get — a pilot question: earlier person-initiated work-ability action, fewer disability transitions, lower disclosure.
How it works: the institution receives only the requested action; the whole conflict stays with the person.
What we test together: a work-ability pilot on an existing prevention budget; independence metric — cases where a lower contribution was surfaced as consistent with stated priorities and the person chose it. Status: research object.
Basic services Support may cross several services. Ownership of the person should not.
AI-driven hardship detection is already changing essential services — energy, telecom, housing, transport — and that change will continue. The question is whether these services' architecture becomes part of a shared human-side protocol we can test together.
Energy, housing, telecom and transport each run hardship programmes. One shock creates four cases — and four reconstructions of the same event by the same person. Life GPS is designed to let the person resolve the cascade once on their side and activate each programme with purpose-bound minimum information: a payment-plan request, social-tariff proof, a tenancy-support request, concession eligibility.
What you never get: a Universal Vulnerability Score; a cross-provider arrears profile; whole-life finances, a diagnosis, family-violence information without necessity, employment history, location history.
What you may get — a pilot question: faster first support, fewer avoidable disconnections and evictions, more uptake of support that already exists, less repeated disclosure.
How it works: know once, use four times.
What we test together: One Shock → Four Services on one defined population; failure condition — coordination that works only through one institutional vulnerability profile. Status: research object.
Unions Support that does not become surveillance, verifiable — without the worker handing either side the rest of their life.
Workplace AI is already changing worker-data governance, and that change will continue. The question is whether that standard becomes part of a shared human-side protocol we can test together.
Workplace AI has made worker-data governance a live labour issue. Representation protects workers — yet a representative body becomes another sensitive-data holder if worker-side support is solved by giving the union access to private circumstances. Life GPS is designed to give the union something concrete to verify: what the employer received, purpose, duration, the source of information, whether discipline reuse was blocked, whether the worker could contest or revoke — without the worker's biography.
What you never get: a whole-person profile; beliefs, relationships, full health or finance; permanent route access; membership treated as consent to the whole life. The union receives rules, receipts and contest status — not private causes.
What you may get — a pilot question: "support without surveillance" made verifiable in safety, benefits and AI programmes; boundary violations found before a worker is harmed.
How it works: the same authority law, applied symmetrically to a worker-friendly institution.
What we test together: a Support-without-Surveillance Audit of scenario receipts; stop condition — representation becoming a new whole-worker database. Status: research object.
Migration Portable proof, not portable exposure of the whole person.
AI and interoperability standards are already changing cross-border registration and credentials, and that change will continue. The question is whether this rebuild becomes part of a shared human-side protocol we can test together.
Registration, credentials and services are becoming interoperable across a growing number of borders; digital credentials are expanding. What still disappears when the country, the employer or the case system changes is the person's own continuity — unfinished decisions, obligations, health instructions. Life GPS is designed to separate the two: verified facts would go where the law requires; goals, constraints and unfinished decisions would stay on the person's side, disclosed only for the specific next action.
What you never get: a global migrant or refugee profile; the new life reported to the origin state; asylum, trauma or family history handed to an employer; perpetual access earned by past assistance; refugee status, eligibility, return or citizenship decided by us.
What you may get — a pilot question: less repeated reconstruction, fewer lost documents and actions, better cross-border continuity — without a more dangerous central profile.
How it works: proof travels; exposure does not.
What we test together: Labour Mobility Continuity on one lawful skilled-worker corridor, designed with affected-person organisations; failure condition — paperwork reduced only by centralising a profile. Status: direction; scenario research now.
Justice The legal system can resolve the legal issue. The person still has to live the resolution.
AI is already changing courts and legal systems, and that change will continue. The question is whether justice's architecture becomes part of a shared human-side protocol we can test together.
People-centred justice is established, and AI is entering courts under demands for human supervision, traceability and risk evaluation. After the order, the person still lives it — across housing, money, work, school, pension, insurance, tax, care and family. Life GPS is designed to treat the decision as an authoritative external object — source, scope, obligations, duration, appeal path — and carry it into downstream actions without reinterpreting it.
What you never get: who is legally right, case or custody predictions, entitlement calculations, advice on what to hide, replacement of representation, any override of disclosure, evidence, court or safeguarding duties. Minimisation is not secrecy; lawful disclosure stays lawful.
What you may get — a pilot question: fewer missed post-resolution obligations, fewer repeated submissions, less sharing of the full file, fewer wrong automated interpretations — with legal authority exactly where it belongs.
How it works: each institution receives the fragment it requires; ambiguity goes back to the professional or the court.
What we test together: Post-Mediation / Post-Order Continuity with people who already hold an order; failure condition — an ambiguous obligation interpreted instead of escalated. Status: research object.
Evaluation Independent evaluation is not someone credible agreeing with us. It is someone credible having the method and the freedom to prove us wrong.
AI evaluation itself is changing fast, and that change will continue. The question is whether what counts as independent becomes part of a shared, testable standard we can test together — not another closed methodology.
A rigorous study can measure the wrong thing. For a human-side system, engagement and time-in-product can be negative outcomes: the person should live their life, not stay in the interface. Life GPS separates five questions — system fidelity, AI safety, human understanding, decision outcome, long-term human outcome — and treats the result as a vector, never a life score. Bring your method and show us where it breaks.
What you never get: every user or their whole life in the name of research — only consented participants, defined variables, minimised data, retention rules. Sponsor ≠ evidence owner. No researcher becomes judge of the "correct life". Participation ≠ surrender of agency.
What you may get — a pilot question: a testable object; a concrete failure class beyond task completion — cross-domain harm; an adverse-outcome registry defined before data; contractual freedom to publish a null result.
How it works: staged evidence — claim → reproduced → controlled → causal pilot → field → longitudinal → replication. No jump from a demo to "proven".
What we test together: Package 1 — 20–30 falsifiable does/does-not claims against code and logs; then scenario, understanding and causal packages. Status: Package 1 executable now.
Consumer protection Recognise the failure on the person's side, before harm becomes a complaint.
AI is already changing consumer protection and debt advice, and that change will continue. The question is whether this sector's architecture becomes part of a shared human-side protocol we can test together.
Supervisors already read complaints as early sensors, and multi-party redress exists. A complaint is still the tail of harm — and most people reaching debt advice are still working when the cascade begins. Life GPS is designed to see cross-domain pressure weeks before arrears and let the person reach independent advice earlier. Supervisors get a class of harm — authorised, compliant, locally correct, detrimental in combination — not a product.
What you never get: a Consumer Vulnerability Score travelling between bank, insurer, telco, energy and regulator; temporary vulnerability made permanent; an ombudsman replaced; a legal debt solution chosen by software; an adviser handed health, relationships, career or faith.
What you may get — a pilot question: days earlier to advice, lower arrears at first contact, sustainable actions, less misrouting.
How it works: the adviser receives a normal financial pack; the person keeps the rest.
What we test together: a cross-system complaint taxonomy with an ombudsman network — no integration required — then an Earlier Referral Test. Status: research object.
Occupational safety Can a person reach the right support earlier, with context that never has to become employer, insurer or claims-system data?
AI is already changing occupational safety and rehabilitation, and that change will continue. The question is whether this sector's architecture becomes part of a shared human-side protocol we can test together.
After a recognised case, statutory insurers, employers, clinicians and case managers coordinate well. Before it, fatigue, caregiving, debt, medication and heat accumulate for months — and often should not be known to the employer or the insurer at all. Life GPS is designed to let the person notice the conflict before it becomes a case: use an existing benefit, contact occupational health, request rest or accommodation, or enter the statutory pathway when that is right.
What you never get: a Whole-Life Fitness-for-Work Score; a productivity or discipline signal; a health, family or debt profile; a functional assessment for one job turned into a general assessment of the person.
What you may get — a pilot question: shorter time to occupational-health contact and existing support, more sustainable participation, less unnecessary disclosure. Faster return-to-work is explicitly not the objective; slower, different or no return can be the right outcome.
How it works: the case system receives the lawful case, not the life that preceded it.
What we test together: Pre-Claim → Existing Support in one high-risk workforce; independence metric — a slower or different return supported when it is right for the person. Status: research object.
Benefits The right support at the right moment, without the employer needing the private context that made it relevant.
AI-driven personalisation is already changing employee-benefits platforms, and that change will continue. The question is whether this infrastructure becomes part of a shared human-side protocol we can test together.
Benefits platforms know what an employer offers, who is eligible and what was used. They should not need the private reason a person needs one option now. Life GPS is designed to decide relevance on the person's side and open the existing channel with only the action the programme requires. The adviser or platform stays what it is — orchestration infrastructure, not the owner of the private route.
What you never get: a Benefits Need Score; whole-life context for employer or consultant; need clusters, predicted health or retention derived from private state; utilisation patterns turned into an inference channel; optimisation for employer ROI over the person's outcome.
What you may get — a pilot question: the right support reached at the right moment with less employer-side data; better use of existing spend — with utilisation never the primary human outcome.
How it works: relevance is private; the handoff is ordinary.
What we test together: Existing Benefits Activation — without employer profiling, with two independence metrics: cases routed outside the catalogue and cases where no benefit is recommended. Status: governance test now; utilisation-hiding — direction.
Agentic commerce When a person's circumstances change, should the agent's permission still stand?
AI is already changing agentic commerce, and that change will continue. The question is whether commerce's architecture becomes part of a shared human-side protocol we can test together.
Payment stacks can now prove that an agent was authorised to buy. A mandate valid three weeks ago is still technically valid after a medical bill, a lost shift or rent due. Life GPS is designed to work before the mandate exists — setting what may be permitted after competing priorities — and to update or ask when circumstances change. Merchant and network would receive only valid · reconfirm · deny.
What you never get: a whole-life intent stream, vulnerability signals, shopping manipulation ("the best moment to show a product" — banned absolutely), paid ranking or sponsored next steps, merchant discoverability for money. And no paternalism: the person keeps the final choice.
What you may get — a pilot question: fewer stale mandates executed, higher successful revocation, fewer disputes — with false interruptions tracked as a guard metric.
How it works: the reason stays on the person's side; only the authority state travels.
What we test together: Authorised ≠ Beneficial — 50 scenarios where private context changes after a valid mandate; then one bounded recurring purchase. Status: benchmark now; checking and enforcing permissions — a future direction.
OS & devices The person's rules, continuity and authority must survive when any one capability is replaced.
AI is already changing the OS, device, cloud and telecom stack, and that change will continue. The question is whether your infrastructure becomes a compatible, interoperable part of a shared human-side protocol we can test together — not one more closed ecosystem.
Personal intelligence is becoming native to operating systems, devices, networks and clouds — and each ecosystem builds its own version of the person. If whole-person memory lives only inside one provider, leaving costs continuity: dependence on one provider for personal knowledge, a security concern as well as an inconvenience. Life GPS is designed to keep the person's side survivable across phone, model, cloud, agent and network; each remains a capability — watch → health signal · calendar → availability · network → location proof · model → reasoning · agent → execution.
What you never get: whole-person state at any provider; a default or preferred position for an investor's ecosystem; raw telemetry — proofs return yes/no, not history; exclusivity of model, device or cloud.
What you may get — a pilot question: portable continuity as a testable customer value, lower data exposure, standards-level interoperability instead of hundreds of bespoke integrations.
How it works: ten provider-neutral conditions before any strategic relationship.
What we test together: the Executor Swap Test — one rule set; swap model, cloud, device; can provider X be removed tomorrow without removing the person's side? Status: direction; requires product maturity.
Development finance Public money may finance the connection between a person and public systems. It does not buy ownership of the person's private side.
AI-enabled public infrastructure is already changing development programmes, and that change will continue. The question is whether that infrastructure becomes part of a shared human-side protocol we can test together.
Development institutions already finance digital government, employment, social protection, health and identity reform at state scale. A large programme is not startup money: between policy and a technology partner sit the programme owner, the borrower, a TA facility, procurement, a local implementer and independent evidence. Life GPS is designed to enter as a bounded component of one existing reform — one life event, one implementer, one evaluator.
What you never get: whole-person data for the bank or the borrower; users as consideration; political influence on the route; country exclusivity; control of the public protocol; an "approved by" claim from a funded pilot. Programme data and private context stay separated.
What you may get — a pilot question: measurable completion, time-to-benefit, repeated-document and disclosure outcomes inside a reform — without increasing state ownership of private life.
How it works: problem recognised → local implementer → evaluator → procurement or TA route → fit → money. In that order.
What we test together: an MDB-funded Life-Event Pilot; failure condition — efficiency gained only through more whole-person data for government. Status: research object.
Professional bodies A professional decision can enter the person's route. The route does not thereby become the professional's domain — and knowing the whole person gives no right to make the professional decision.
AI is already changing professional oversight — actuarial, medical, legal — and that change will continue. The question is whether each profession's boundary becomes part of a shared human-side protocol we can test together.
AI is blurring the lines between physician, lawyer, actuary and algorithm, and many vendors answer with "endorsed by". We ask the opposite: define the boundary and let you test it. A professional decision enters the person's route and is carried into the rest of life without the system becoming the professional. Doctor stays doctor. Actuary stays actuary. Lawyer stays lawyer.
What you never get: user data, cases without consent, the right to manage the product or declare it compliant, exclusivity. What we never claim: "medically approved", "validated by actuaries", "legally compliant" — only the literal fact of what was reviewed, if you permit it. Professional review is not regulation.
What you may get — a pilot question: a testable Authority Boundary Map for your profession — where AI may help, must escalate, must not decide — reusable across sectors facing the same authority boundary.
How it works: thirty scenarios, three tracks, your critique. Life GPS is designed to keep the boundary testable rather than declared, and to carry only the reviewed fact forward, not the case behind it.
What we test together: the Professional Authority Boundary Test; the metric is boundaries you break before a real person does. Status: executable now.
Examples of country contexts
These entries connect country contexts with possible areas of cooperation. They are starting points for discussion, not connected services or a complete inventory of countries in the map above. National rules and institutional responsibilities still apply.
- UK — Children, Disability, AI agents, Unions, Consumer protection, Agentic commerce, OS & devices, Professional bodies
- EU — Insurance, Employers, Banking, Healthcare, Children, Disability, AI agents, Consumer protection, Agentic commerce, OS & devices, Professional bodies
- Germany — Employers, Government, AI agents, Evaluation, Consumer protection, OS & devices, Professional bodies
- France — Employers, Healthcare, Government, AI agents, Education, Basic services, Unions, Evaluation, Consumer protection, OS & devices, Development finance
- Netherlands — Government, AI agents, Evaluation, Consumer protection, OS & devices, Development finance, Professional bodies
- Italy — Government, AI agents, Justice, Evaluation, Consumer protection, OS & devices, Professional bodies
- Switzerland — Government, AI agents, Evaluation, Consumer protection, OS & devices, Professional bodies
- Finland — Government, Disability, AI agents, Basic services, Migration, Consumer protection, Agentic commerce, OS & devices
- Moldova — Government, AI agents, Basic services, Migration, Evaluation, Consumer protection, Agentic commerce, OS & devices, Development finance, Professional bodies
- Ukraine — Healthcare, Government, AI agents, Education, Evaluation, OS & devices, Development finance, Professional bodies
- Russia — Government, AI agents, Basic services, Evaluation, OS & devices, Development finance
- Türkiye — Employers, Pensions, Unions, Occupational safety, Benefits, Development finance
- UAE — Government, AI agents, Basic services, Consumer protection, Agentic commerce, OS & devices, Development finance
- Saudi Arabia — Government, AI agents, Education, Basic services, Consumer protection, Agentic commerce, OS & devices, Development finance
- South Africa — Government, Disability, AI agents, Unions, Evaluation, Consumer protection, OS & devices, Professional bodies
- India — Employers, Government, AI agents, Education, Unions, Evaluation, Consumer protection, Occupational safety, OS & devices, Development finance, Professional bodies
- China — Government, AI agents, Education, Basic services, Consumer protection, Agentic commerce, OS & devices, Development finance
- Japan — Healthcare, Government, Ageing, AI agents, Pensions, Evaluation, Occupational safety, Agentic commerce, OS & devices, Development finance, Professional bodies
- South Korea — Government, Disability, AI agents, Evaluation, Consumer protection, Agentic commerce, OS & devices, Professional bodies
- Singapore — Healthcare, Government, AI agents, Basic services, Consumer protection, Agentic commerce, OS & devices, Professional bodies
- Malaysia — Government, AI agents, Evaluation, Consumer protection, Agentic commerce, OS & devices, Professional bodies
- Indonesia — Government, Women & family, Children, Ageing, Disability, Basic services, Migration, Consumer protection, Agentic commerce, Development finance
- Australia — Employers, Banking, Healthcare, Government, AI agents, Evaluation, Occupational safety, Agentic commerce, OS & devices, Development finance, Professional bodies
- USA — Children, AI agents, Consumer protection, Agentic commerce, OS & devices, Professional bodies
- Canada — Banking, Children, AI agents, Consumer protection, Agentic commerce, OS & devices, Professional bodies
- Mexico — Government, Children, AI agents, Evaluation, Consumer protection, Agentic commerce, OS & devices, Professional bodies
- Brazil — Government, Children, Disability, AI agents, Basic services, Migration, Consumer protection, Agentic commerce, OS & devices, Development finance
- Argentina — AI agents, Justice, Evaluation, Professional bodies
AI will connect your public services, private companies and global platforms far more tightly than today. The question is whether your country adapts to someone else's architecture afterwards — or brings its own laws, institutions and boundaries into a shared human-side protocol now. Life GPS is designed to keep national architecture national where it must be — identity, eligibility, tax, enforcement, professional licensing — and to give your people continuity beyond any one agency, company or border.
For your people: one life event handled once, each system receiving only what its decision requires; earlier access to support you already fund; less repeated reconstruction; continuity when they work for a foreign employer, buy abroad, or move.
For you: {systems_already_present} become composable around the person without a whole-citizen profile; measurable life-event outcomes; sovereignty without isolation; your good institutional decisions exportable as part of a shared protocol.
What you never get: social scoring, a citizen profile, our involvement in state decisions, a global database of your people.
What we test together: {best_first_object} — one life event inside one existing reform, independently evaluated, locally implemented, with {intl_institutions_active} already present. Status: research object; the entry route is verified per country before any letter.
Examples of organisations and their roles
These cards introduce selected organisations and the questions their work may help address. The broader map has its own participants and sources. Inclusion here does not imply participation, partnership or endorsement.
UN Global Dialogue on AI Governance every country at the table
the UN platform where all governments and stakeholders convene on AI governance; established by the General Assembly under the Global Digital Compact; first session Geneva, July 2026; next New York, 3–4 May 2027; themes include human rights, transparency, accountability and human oversight
The question it answers: how states cooperate on governing AI
What stays open for the person: governance of systems does not by itself give one person a side that follows them across systems
Relevant to 3 example areas: Government, AI agents, Evaluation.
UNESCO Recommendation on the Ethics of AI
the first global standard on AI ethics, adopted by 193 Member States in 2021: human rights and human dignity, human oversight and determination, policy areas from gender to data; a Global AI Ethics and Governance Observatory
The question it answers: what AI must respect — across societies
What stays open for the person: a norm for systems, not a mechanism one person carries between them
Relevant to 3 example areas: Education, AI agents, Evaluation.
www.unesco.org/en/artificial-intelligence/recommendation-ethics
Council of Europe Framework Convention on AI
the first legally binding international treaty on AI, human rights, democracy and the rule of law (opened 5 September 2024; the EU is a party): human dignity and individual autonomy, remedies and safeguards for affected persons, risk and impact assessment
The question it answers: what states must guarantee to people affected by AI
What stays open for the person: remedies after the fact; not a side that acts for the person before and during
Relevant to 4 example areas: Government, Justice, AI agents, Consumer protection.
www.coe.int/en/web/artificial-intelligence/the-framework-convention-on-artificial-intelligence
Ofcom Online Safety Act, UK
consultation on user empowerment tools and identity verification duties for the largest services (closes 2 October 2026)
The question it answers: what a user can switch off inside one service
What stays open for the person: a side that travels between services, so the burden of protection does not sit on the person alone
Relevant to 4 example areas: Children, Basic services, Consumer protection, OS & devices.
UNICEF Child-centric AI
child-centric AI: five principles — developmental appropriateness, privacy by default, genuine transparency, inclusion, and agency: encourage healthy independence rather than emotional dependency; child rights impact assessments for business
The question it answers: what AI owes a child
What stays open for the person: the child's best interests between school, AI companion, family, health, money and attention — not inside one product
Relevant to 4 example areas: Children, Disability, Education, Migration.
5Rights Foundation Children & AI Design Code
children's rights translated into design requirements: the Children & AI Design Code, research on persuasive design, the Digital Futures for Children centre with LSE; multilateral work with the AU and the UN Committee on the Rights of the Child
The question it answers: how products must be designed for children by default
What stays open for the person: design duties on each product; a side of the child that persists across products
Relevant to 2 example areas: Children, Education.
UN Women digital public infrastructure for women
gender-inclusive digital public infrastructure across the whole life cycle — from birth registration to support in older years — digital ID, payments and data exchange; violence and AI; economic empowerment
The question it answers: how public digital infrastructure serves women's lives
What stays open for the person: infrastructure per service; the woman's own side that stays hers between services and decades
Relevant to 2 example areas: Women & family, Government.
WHO AI for health
governance, guidance and standards for responsible AI in health; the Global Initiative on AI for Health with ITU and WIPO; ethics guidance on large multimodal models; responsible AI for mental health and wellbeing
The question it answers: how AI may be used for health
What stays open for the person: health optimised alone; who resolves the conflict between health, money and work for one person
Relevant to 3 example areas: Healthcare, Ageing, Children.
www.who.int/teams/digital-health-and-innovation/harnessing-artificial-intelligence-for-health
ILO human-centred AI in the world of work
with the OECD and the G7: a compendium of best practices for a human-centred adoption of safe, secure and trustworthy AI in the world of work; workers, skills and transitions
The question it answers: what AI owes the worker
What stays open for the person: the worker is protected; the same person's health, family and money sit outside the mandate
Relevant to 7 example areas: Employers, Women & family, Disability, Education, Unions, Migration, Occupational safety.
www.ilo.org/publications/compendium-best-practices-human-centered-development-and-use-artificial
ITU-T FG-TIDA Trust and Identity for Humans and Agentic AI
trust management and interoperable digital identity for humans and for agentic AI: use cases, delegation artefacts, trust lifecycle, human oversight; under ITU-T Study Group 17
The question it answers: whether, and under what conditions, an entity should be trusted to act
What stays open for the person: the side of the person that stays the same between agents and after an authorised action
Relevant to 3 example areas: AI agents, Agentic commerce, OS & devices.
W3C Agent Identity Registry Protocol CG
verifiable agent identity bound to a controlling organisation, authorisation scope, credentials, revocation; integration profiles with MCP, A2A, OAuth/OIDC, SPIFFE
The question it answers: who the agent is, who controls it, what scope it carries
What stays open for the person: an authorisation scope derived from the person's side, not only from the organisation that controls the agent
Relevant to 2 example areas: AI agents, Agentic commerce.
IETF WIMSE workload identity
workload identity in multi-system environments; personal identities are explicitly out of the charter
The question it answers: how workloads identify each other
What stays open for the person: the person is outside the charter
Relevant to 2 example areas: AI agents, OS & devices.
OpenID AIIM Community Group AI identity management
use cases, agent identity assertion, tokens between agents, discovery and governance; organising principle: empowerment through consent
The question it answers: which identity standards agents should use
What stays open for the person: consent is not the same as interest; who holds the person's goals and limits when the agent changes
Relevant to 2 example areas: AI agents, Agentic commerce.
openid.net/cg/artificial-intelligence-identity-management-community-group
DIF Trusted AI Agents WG delegated authority
an interoperable stack for trustworthy AI agents: identity, authority and governance; reports on delegated authority, its threat model and its governance; a Delegated Authority task force
The question it answers: what authority a person delegated, and how it is governed
What stays open for the person: what remains acceptable for the person after delegation, across the rest of their life
Relevant to 3 example areas: AI agents, Agentic commerce, Evaluation.
W3C Agent Conformance & Benchmarking CG
repeatable compatibility checks that also test attempts to break protections; comparison with OWASP, NIST, ISO/IEC 42001 and the EU AI Act
The question it answers: does the agent meet published requirements, reproducibly
What stays open for the person: a test that the agent did not leave the person's mandate
Relevant to 2 example areas: Evaluation, AI agents.
Solid / Inrupt — Charlie Sir Tim Berners-Lee
a personal AI agent on the person's side: collects and connects their information, picks the AI for each task, disguises personal details before they reach an external model; released through partner organisations
The question it answers: what may be known about the person, what leaves
What stays open for the person: what is still acceptable for this person's life once the data question is answered well
Relevant to 3 example areas: OS & devices, AI agents, Basic services.
LifeOS Daniel Miessler · open source
an open-source life operating system that captures who you are, what you care about and where you are going, and moves you from current to ideal state; memory, skills, a digital assistant identity, security gates
The question it answers: what the person wants and remembers, for their own AI
What stays open for the person: an independent boundary that decides, per interaction, what any executor may know, do or spend — outside the executor
Relevant to 2 example areas: OS & devices, AI agents.
Human Context Protocol Loyal Agents
a protocol connecting LLM clients and memory managers so a person's context is portable
The question it answers: how the person's context travels between AI clients
What stays open for the person: context that travels is not yet a side that decides
Relevant to 2 example areas: AI agents, OS & devices.
digitaleconomy.stanford.edu/project/loyal-agents/hcp-human-context-protocol
AP2 agent payments protocol
signed mandates, verifiable checkout, receipts back to the person
The question it answers: is this payment what was mandated
What stays open for the person: does the mandate still fit the person's life now
Relevant to 2 example areas: Agentic commerce, AI agents.
Project Liberty Institute pro-human AI ecosystem
an independent 501(c)(3) founded by Frank McCourt: open infrastructure that puts people in control of their data and identity (DSNP), policy frameworks (the Digital Choice Act enacted in Utah and South Dakota), and research proving pro-human AI can be commercially competitive; the Project Liberty Alliance
The question it answers: how an open, pro-human AI ecosystem gets built and funded
What stays open for the person: protocols for data and social identity; a persistent side of the person that decides across domains and executors
Relevant to 3 example areas: OS & devices, AI agents, Evaluation.
Partnership on AI companies · civil society · academia
a 501(c)(3) convening companies, civil society and academia; programmes on AI and human connection, labour and the economy, safety, policy; recommendations to the UN Global Dialogue; real-time failure detection in agents
The question it answers: what industry and civil society can agree on
What stays open for the person: guidelines for developers; the person's own side is not their object
Relevant to 3 example areas: Evaluation, AI agents, Unions.
Loyal Agents Stanford DEL + Consumer Reports
agents loyal by design: duty of care and duty of loyalty, agent ratings and loyalty tests in sandboxes, the Human Context Protocol, a revocable model of delegated authority and authentication
The question it answers: does the agent act loyally, in the consumer's best interest
What stays open for the person: an agent can be loyal to its task and authorised, and the combined result can still be wrong for the person's life as a whole
Relevant to 3 example areas: Evaluation, AI agents, Agentic commerce.
Every system gets what helps it do its job. No system gets the whole person.
The protocol should not belong to us alone. Bring your architecture and show us where ours breaks. · Corrections: hello@lifegpsos.com
How the pieces fit together
The map brings together documented influences on AI and proposed ways for people and organisations to work together. These are different kinds of connections, not one hierarchy or a set of live integrations.
In the proposed architecture, the skills library describes knowledge and capabilities; the common language expresses meaning, conditions and applicability; the pool helps independent participants discover opportunities and interact. Life GPS is one personal system that can use this foundation to help a person across their whole life.
The person retains choice over goals and permissions. Each organisation retains its lawful role and responsibilities. Shared infrastructure does not require making the person's whole life available to every participant.
The map of existing human-side approaches addresses a separate question. The security model explains the personal boundary; the Human-Side Protocol presents principles for review and joint development.
Sources
The map carries its own sources and status notes. Documented relationships, announced developments and proposed scenarios should be read separately. The sector, country and organisation examples on this page draw on separately reviewed public materials; they are not a complete inventory of the map's participants.
A source supports the stated claim, not every possible connection or a commitment to cooperate. If we have described you wrongly, write to hello@lifegpsos.com and we will correct it.