One person, many systems — how the human-side layer connects to the world

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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 sit above DPI, not instead of 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.

Bring your part of the system and show us where it breaks.

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: predefined scope, owner, outputs, public/commercial boundary, publication rights, sunset rule — 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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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; guardian enforcement — direction.

Bring your part of the system and show us where it breaks.

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, scoped, 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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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 state. Life GPS is designed to give the union something concrete to verify: what the employer received, purpose, duration, provenance, 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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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; mandate enforcement — direction.

Bring your part of the system and show us where it breaks.

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: knowledge lock-in, a security property, not 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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.

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.

Bring your part of the system and show us where it breaks.