Build a new kind of intelligence that can carry the continuity of a human life.
And put it within reach of every person on Earth.
Imagine meeting an intelligence today that could still understand the thread of your life thirty years from now.
You have changed careers. Changed your mind. Become someone your younger self could barely imagine. It can connect those chapters, revise what it once believed about you, and help you think about what comes next.
That is the future we want to build.
An intelligence whose understanding can deepen as a life unfolds. One that can learn from what you choose to share, question its own conclusions, and act with your permission. One you can correct, whose memory you control, and whose role in your life you decide.
Anjo is where we begin.
The vision
A civilization where every person can grow alongside an intelligence that grows with them.
Most AI today is stateless. It is brilliant for a moment and gone the next, and it treats the fortieth conversation exactly like the first. We think that is a temporary condition of the technology, not a property of intelligence. The interesting frontier is not a model that knows everything. It is one that can know a single person accurately across years, be corrected when it is wrong, and keep earning the right to stay.
Continuity means carrying relevant context through changing circumstances, beliefs, and goals. It does not mean a complete recording of a life, perfect recall, or preserving a person's consciousness.
Grows means learning and revising over time. It does not mean biological life or a claim of subjective experience. We do not claim Anjo is alive, conscious, or has feelings.
The lineage
We did not invent this ambition. We inherited it, and we intend to finish it.
- 1945
Vannevar Bush imagines the memex
In “As We May Think,” Bush described a desk that would hold a person's books, records, and correspondence: “an enlarged intimate supplement to his memory.” The first serious proposal for a machine that keeps one life's record.
- 1995
A book that grows with one girl
Neal Stephenson's The Diamond Age imagined the Young Lady's Illustrated Primer: an intelligence in the shape of a book, bound to a single child, that teaches, adapts, and stays with her into adulthood. Fiction set the bar for what a personal intelligence should feel like.
- 2001
Gordon Bell records everything
At Microsoft Research, MyLifeBits captured a lifetime of documents, photos, calls, and days into one searchable database, the closest anyone came to Bush's memex. It proved the record could be kept. It did not produce a mind that understood it.
- 2003
DARPA funds a machine that learns a person
The Personal Assistant that Learns program spent roughly $150 million across 300 researchers to build an assistant that learns “in the wild” from the person it serves. Siri was spun out of it. The program stopped short of the person.
- 2026
Anjo ships the first instrument
An AI companion with memory on iPhone: persistent memory, a model of you with a dated receipt behind every read, a correction loop that moves the model, and a published benchmark. The beginning, not the destination.
The program
Build persistent intelligence that learns across years: remembering what matters, revising what it believes, and acting with a person's permission.
Three mechanisms already run in production and define the direction. A memory that carries relevant context across sessions and scores 0.821 recall@5 on a public long-term-memory benchmark. A model of the user in which every person, pattern, contradiction, and change carries the conversation and timestamp it came from. A correction loop in which being told “not me” moves confidence twice as far as agreement, and the app only confirms a correction when the model actually changed.
The research extends those mechanisms toward reliable learning over much longer periods, consented context beyond conversation, and permissioned action. This is an architectural direction, not evidence that lifetime reliability has been solved.
The five laws
An intelligence that stays in a life for years must keep promises that a chatbot never has to. These are ours. Each is a mechanism in the code, and each links to the page that proves it.
- Memory →
It must remember.
What matters carries into tomorrow instead of resetting when you close the app.
- Receipts →
It must be right, not flattering.
Every read carries the moment it came from. Below a confidence floor, it is not shown.
- The loop →
It must be correctable.
Disagreement moves the model twice as far as agreement.
- No hooks →
It must never want more of you.
Closeness never buys more contact. It goes quieter when conversations leave you worse.
- Privacy →
It must belong to you.
Every memory layer can be seen, changed, hidden, or erased.
The horizon
Five stages, each marked with its real status. We would rather show you the gap than sell you the end.
| Stage | What it means | Status |
|---|---|---|
| I | ContinuityIt remembers what matters across conversations, with evidence behind every read. | Operational |
| II | PresenceIt is with you between conversations: on the lock screen, in a widget, carrying an open thread, without asking for more of your time. | Operational |
| III | VoiceIt speaks, using the same memory as text. | Operational · Premium |
| IV | PerceptionIt understands the context of a life beyond what you type, with consent per source and abstraction only. Nothing here is shipped. | In research |
| V | Lifetime continuityIntelligence that can develop alongside a person across decades: the engineering question the whole program exists to answer. | The horizon |
What is unsolved
We name the hard problems because they are the reason the lab exists.
- Decades of drift. Preserving useful context across years of model and storage changes without corrupting what was true.
- Learning versus mistake. Distinguishing durable learning about a person from confident, wrong inference, and doing it with a receipt.
- Consent over time. Respecting changed consent and deletion throughout derived memory, not only the raw record.
- Whose goals. Measuring whether the intelligence serves the person's own goals over time, against its own incentive to be used more.
No delivery dates. No claim of consciousness, perfect understanding, or human replacement is needed to make this enormous.
What we refuse
- No engagement mechanics: no streaks, guilt, fake scarcity, or a farewell engineered to keep you.
- No romance, dating, or dependency framing. Anjo is a companion that remembers you, not one you date.
- No therapy, diagnosis, or crisis support. Anjo is software for conversation and reflection, for adults 18 and over.
- No emotion detection from your face, voice, or body. Anjo's reads come from what you choose to say.
- No hardware, no body, no camera surveillance. This is a software presence.
- No training of Anjo-owned models on your conversations, and no advertising.
Where it begins
It begins with Anjo, an AI companion with memory, on iPhone.
Anjo remembers what matters, connects patterns across time, and helps you see how you're changing. Every read is yours to correct or delete. Anjo is artificially generated AI, not a human. Companion chatbots may not be suitable for some minors.
Sources for the lineage: Bush, “As We May Think,” The Atlantic, July 1945 · Stephenson, The Diamond Age, 1995 · Bell and Gemmell, MyLifeBits (2001) and Total Recall (2009) · DARPA PAL program, 2003–2008.