Pre-launch · archive in design

Everything about your life sits in accounts that end when you do.

EternalOS is building a durable record of a person — memories, voice, writing, documents, the reasoning behind decisions — described and structured so it can still be read in a century. Not a copy of a mind. A record that survives the platforms it was made on.

  • Open formats only
  • Several independent copies
  • No simulation of the person
The problem

The archive you think you have

Three reasons the record of a life is far less durable than the amount of it suggests.

It is in somebody else's account

Photos, messages, documents and everything else live under terms of service, tied to a login, in companies with their own lifespans. Access ends with the account, and the account ends with the person or with the company.

Formats stop opening

The proprietary format that everything was saved in is readable until the software that made it is gone. Anyone who has tried to open a family video from 2005 already knows how this ends.

Volume is not a record

A terabyte of undescribed photos is not a memory; it is a haystack. Without knowing who is in them, when, where and why they mattered, the quantity works against whoever comes looking.

How it works

Capture, structure, keep

Three stages, and the third is the one the whole thing is actually about.

  1. 01

    Capture

    Recorded conversations, writing, correspondence, documents and images — gathered with the person deciding what belongs in the record and what does not. Selection is part of the work, not a compromise on it.

  2. 02

    Structure

    Every item described: who, when, where, in what context, and how it relates to everything else. An indexed, navigable record rather than a folder of files somebody will one day have to guess at.

  3. 03

    Keep

    Open formats, several independent copies in different places, checked and migrated on a schedule, with a written plan for what happens to the archive when the company holding it no longer exists.

How it is built

Six decisions the archive rests on

Preservation is not a feature you add later. These are the choices being made first, because they are the ones that cannot be changed afterwards.

Open formats, no exceptions

Documented, non-proprietary formats with independent implementations, for everything. If a format needs one particular company's software to open it, it does not go in.

Description travels with the content

Metadata and provenance embedded alongside the material rather than kept in a database that could be separated from it. A file without its description is an orphan in fifty years.

Several copies, several holders

Independent copies in different places and under different legal jurisdictions, verified against each other on a schedule. One copy is not preservation, it is storage.

Access rules recorded with the record

Who may see what, and from when — decided by the person while they are here to decide it, and stored as part of the archive rather than as a setting in an interface.

A plan for outliving us

A defined route by which the archive passes to an institution built for permanence if this company stops. Written before the first record exists, because afterwards is too late.

Readable without us

A plain, self-describing index that somebody can navigate with no account, no application and no knowledge of how any of this was built.

What this is not

This field attracts claims that cannot be supported, and the products being sold today are mostly simulations trained on somebody who is gone. We are not building that, and the distinction is the reason the project exists rather than a caveat attached to it.

  • Not a mind upload. Nothing here copies, transfers or preserves a consciousness. It is a record, in the sense a library uses the word.
  • Not a simulated person. No chatbot of your grandmother. A record answers with what was actually said; a simulation answers with something plausible, which is worse than silence.
  • Not immortality. The person is not preserved. What they wrote, said, made and decided can be, and that is a different and achievable thing.
  • Not a medical or spiritual claim. No statement about death, continuity or what happens afterwards is being made anywhere in this project.

The long horizon

There is a design question underneath all of this: what must a record contain to be understandable by someone with none of its context — no company, no shared language, no living witness? The most extreme version of that question is a record sent beyond Earth, and using it as a constraint produces better archives here, which is the honest reason it is in the programme at all.

  • A design constraint, not a service. No capability to send anything interstellar exists, ours or anyone else's. Nothing on this site is an offer to do it.
  • The distance is the point. Alpha Centauri is about 4.37 light years away. Any record that could survive that journey has to explain itself completely.
  • Self-describing or nothing. The constraint forces the archive to carry its own key: what the formats are, what the symbols mean, how the pieces relate.
  • The near-term benefit is real. A record built to be read without context is also a record your great-grandchildren can read without us.
Where we are

Nothing is being kept for anybody yet

An archive that has not proven it can preserve anything should be careful about accepting things to preserve.

  1. Now

    Formats and structure

    Deciding what the record actually is: formats, schema, description, and the migration process — tested on our own material, where the consequences of getting it wrong are ours.

  2. Next

    The first complete records

    A small number of people, taken end to end from capture to a verified multi-copy archive, with the process documented publicly enough to be criticised.

  3. Later

    The structure that outlives the company

    The legal and institutional arrangement — a trust, an endowment, a partner archive — that makes the promise credible past our own existence. Nothing about this is solved by good intentions.

FAQ

Frequently asked questions

What is EternalOS in one sentence?

An organisation building durable, described, machine-readable records of individual people, in open formats and multiple copies, designed to remain readable long after the tools that made them are gone.

Is this uploading a mind?

No. Nothing in the project copies or preserves consciousness, and no result from it would bear on whether that is possible. What is preserved is what a person wrote, said, made and decided — the same category of thing an archive has always held, kept better.

Will there be a chatbot of the person?

No, and this is a deliberate position rather than a limitation. A model trained on somebody produces plausible new statements they never made, which corrupts the record it was built from. The archive answers with what was actually said, or says it does not know.

Who can see the archive?

Whoever the person decided, from whenever they decided. Those rules are recorded as part of the archive rather than as a setting we control, and they are meant to be legible to whoever holds it in fifty years.

What happens if the company shuts down?

That is the central design problem, not an edge case. The answer has to be a written succession route to an institution built for permanence, plus copies already held elsewhere by people other than us. Until that exists on paper, we are not asking anyone to trust us with a record.

Is anything actually being sent to Alpha Centauri?

No. There is no capability to do that, ours or anyone else's, and there will not be for a long time. It is used as a design constraint: a record that could survive with none of its context is a better record here. Anyone selling it as a service is selling something that does not exist.

What will it cost?

Not decided, and it would be dishonest to publish a number before the preservation process is proven. What is decided is the shape: preservation costs are ongoing rather than one-off, and any pricing that pretends otherwise is quietly promising something it cannot fund.

Come back when there is something to keep

The waitlist form goes live soon. Until then the blog explains the decisions being made, including the ones that argue against the whole idea.

No form yet, no data collected — just come back.