A personal AI that sits beside you, not on top of you.
LifeOS is a personal chief of staff for a more meaningful life. It plans the day across work, health, relationships and growth, learns from short check-ins, and treats privacy as a design constraint rather than a feature.
AI tools know your calendar. They don't know your life.
General AI assistants lack the context to plan a real day, and they treat life as if it were only work. Habit apps go the other way: you end up living inside the app, ticking boxes, serving the tool instead of your life.
There is a bigger pattern underneath. With AI, one person can do almost everything alone, and many end up isolated: health, friends and connection slip. LifeOS is built to push in the other direction.
Ten minutes a day, at most
The attention budget is a design rule: up to ten minutes on a normal day, three to five on a busy one.
Morning brief
Saying good morning is enough. LifeOS answers with the few things that matter today, across all areas of life.
One-line check-ins
“Did the morning things, skipped the run” replaces ticking ten boxes. The system works out the rest.
Journeys
21 to 30 day paths toward one specific change, such as a morning ritual. One or two active at a time, so focus holds.
Visual frameworks
Wheel of life, Eisenhower matrix and others show where today's effort lands, and help shrink the zones you don't want.
Agents that follow up
Gentle follow-ups and suggestions. Routines are never changed without asking first.
Colour with meaning
Work-shaped areas lean teal, life-shaped areas lean purple. Red only means a real system error, never that you failed.
Built so that a leak would reveal very little
The biggest barrier to an app like this is the worry about privacy. So privacy is built in fully, even though it is not the headline.
Private content, user-held key
Reflections and personal notes are encrypted on the device. The user decides what counts as confidential and sees what the AI can no longer do with it.
Identities kept apart
Names, numbers and places live in a separate store and are replaced by IDs in everything else, including what reaches a model.
Right model, no retention
Heavy reasoning goes to a frontier model through endpoints that don't retain data. Sorting and extraction run on a self-hosted open model.
Store less
Raw text is deleted once what matters has been extracted.
Retention by period
At the end of each quarter the data is handed to the user and deleted from the servers. Only context and lessons carry over.
A pattern never outlives its evidence
What the system concludes about you expires when the data behind it does.
Leaving means leaving
Users who leave get all their data, and everything is deleted except the username and anonymous performance numbers.
Private beta, modular by design
LifeOS is in private beta with its first users. It is built as modules by life area, such as languages, sport and nutrition, so new areas can be added without touching the core. Every release is versioned, data survives updates, and any release can be rolled back.
Next is a two-way integration, so users can connect their own agents and tools to LifeOS and reach their LifeOS data from their own systems.
Building a product that handles personal data?
We design privacy into the architecture from the first sprint. Let's talk about yours.