Life GPS Try the beta

AI should not replace people. It should know when another person is more valuable.

In short

The more capable AI becomes, the easier it is to replace a person with it. Life GPS uses AI for the opposite purpose: to recognise the moments when another human being is more valuable than any model — and to bring that person into the route.

This is not a marketplace of experts, and not a companion that stands in for people. It is a network where help moves between people who are actually on each other's paths, with AI doing the part humans are bad at: seeing what matters to both sides before they talk.

Available today: the relationship structure, the closed network, skills and reciprocity, and the preparation of both sides before a conversation. Built and in testing right now: the live call itself and the mediator — they switch on after the current stage. So today the system brings a person to the point of the conversation; for the moment the conversation itself happens outside it.


The problem this addresses

Two things are happening at once.

People are becoming easier to replace. An AI answers instantly, never gets tired, never judges and costs nothing. For a growing number of situations that is enough — and for a growing number of people it becomes the default. The quiet cost is that the person who could have helped is never asked, and the relationship that would have been built never is.

And help became a market. The knowledge a person needs usually exists a phone call away — someone in their own circle has already been through exactly this. But there is no path from the need to that person, so the need goes to a search box, a subscription, or an AI.


What Life GPS does instead

It knows when a person is worth more than a model

Inside a route, some steps are answered better by AI, some by a service, and some only by another human being — someone who has been through the same thing, or someone who is already part of this person's life and needs to be part of the decision.

The system routes to the third case deliberately. It is not a fallback for when the AI fails; it is a separate, better answer for a class of situations.

It understands relationships as structure, not as a contact list

A network is not a list of names. Parents, children, a partner, siblings, grandparents, the wider family, close friends, old friends, friends scattered across the world, colleagues, people in the same place — each of these is a different kind of relationship with a different path, and the system treats them that way.

In the beta.

It prepares both sides before the conversation

When a step requires a conversation, the system does not just say "talk to them". It prepares each side for what matters to the other — so that the conversation starts from the real thing instead of the fifteen minutes it usually takes to get there.

Preparation is in the beta. The calls themselves are built and in testing.

The mediator: AI between two people, replacing neither

When two people are at cross purposes, the mediator does not relay their words. Knowing what is important to each side, it surfaces what may matter to both — the thing one of them could not say and the other could not hear.

Built, in testing now — switches on after the current stage.

What this does not claim. No clinical or therapeutic effect. The mediator does not replace either person, and it does not replace professional help. It is a way to start a conversation better, nothing more.

Reciprocity instead of commerce

A person helps with what they have already lived through, and gets help with what someone else has. The currency is time, not money. Reciprocity is visible — who helped whom, and with what — without anyone being scored.

People are never rated. No public ratings, no quality numbers on human beings, no featured tier, no way to buy a place in someone's route.

Code—live — skills and reciprocity are in the beta.

The network is closed on purpose

Invite only. No likes, no stories, no follower counts, no feed. Only people who are actually in a position to help — not an audience.

Code—live.


One situation: a family

A person decides to change careers. It affects their income for a year, and it affects everyone living with them.

What a typical product does. Treats it as one person's goal. Builds a plan for that person. The family finds out when the plan is already in motion — which is where most of the real damage is done.

What Life GPS does. It knows this person, and it also knows there are people whose lives move when this one does. So:

The first two and the finding are in the beta. The call inside the system is built and in testing — for now that conversation happens outside it.

Why this is only possible whole. A career app sees the career. A calendar sees the calendar. Neither can see that a decision in one sphere of one person lands in the middle of someone else's week — because neither holds the relationship as part of the context.

Three laws this stands on. People are not signals. AI mediates, it never recommends people as products. Trust outranks ratings. — from the constitution of the network, written before the code.

What "trust" means here, since we also say nobody is rated

Trust in this system is not a score, hidden or otherwise. There is no number attached to a person, nothing that ranks one human above another, and nothing that can be compared between people.

What exists is narrower and situational: whether this relationship is one where help has actually moved, and whether this particular person has been through this particular thing. It belongs to a pair and a context, not to a person — it does not travel with them, cannot be displayed, and cannot be used to sort anyone.

The distinction matters because a "trust score" is exactly what this constraint exists to prevent.


How wide this reaches

Family first, because those are the decisions that move more than one life at a time — a career change, a move, money, someone getting ill, a child growing up faster than the family's arrangements.

Then friendships that quietly go dormant because nobody has a reason to call. Professional circles where the person who solved this exact problem is two contacts away and unreachable. Communities of people going through the same thing at the same time. And love, which has its own area — matched by the shape of a life rather than by a photograph.

Anywhere one person's choice lands in someone else's week, this is the layer that was missing.

Relationships, the closed network, reciprocity and conversation preparation are available in the beta. Live calls and the mediator are built and in testing now — they switch on after the current stage.


What we are not building

Not a marketplace. Nobody buys placement in a person's route, and nobody is ranked for sale.

Not a social network. No feed, no audience, no engagement metrics. The system does not need a person to stay inside it to have succeeded — a short session is a good session.

Not a companion. The system does not offer itself as a substitute for the people in a person's life. That is the exact failure mode this whole system exists to avoid.


Where this connects

To the whole life. The network is not a separate app — it is what happens when a step in the route needs a person → One life. One context. One route.

To the living avatar. The system knows what kind of relationship this is and what has actually happened in it, and that changes over time → A person is not a profile.

To the human antivirus. No ratings, no engagement mechanics, no pressure to respond, and safety routing overrides everything else → The more a system can do, the less it should control.


What you can check today

The iOS beta is open: TestFlight.

What you will see: relationships by type, the closed network, skills and reciprocity, support by sphere, call preparation. What you will not see: live calls and the mediator — built, switched off until the next stage.


Life GPS · lifegpsos.com · published 16.09.2026

Published 2026-09-16 · Updated 2026-09-16 · Every claim on this page carries a status