Typed Memory
Every memory entity carries a category: People, Documents, Facts, Events, Episodes, and further types beneath them. Queryable, editable, versioned, audited.
What AI Twin does
AI Twin is a typed personal operating system that remembers carefully, surfaces what counts, and stays quiet by design.
Running a practice? Set up your own Twin first, then add the practice from the workspace switcher. It sits alongside, kept apart.
Everything is free for the first 7 days. No card needed. Add one when you decide to stay.
Every memory entity carries a category: People, Documents, Facts, Events, Episodes, and further types beneath them. Queryable, editable, versioned, audited.
Nothing is saved until you confirm. Duplicates and conflicts come to you to decide.
Your Twin spots the same person twice, and two facts that disagree. You choose: merge, keep both, or keep the newer one.
See it by room, by person, on a timeline, or all at once.
A map of the people in your life, drawn only from what you confirmed. Take anyone off the map without erasing them.
Typed lookup for precise facts, vector search for meaning, full-text search, and facts that know when they were true, combined into a single answer.
Search across your memory with typed filters and natural language. Cross-Workspace retrieval requires explicit confirmation.
Memory organised across four horizons: recent, intermediate, long-term, archived.
A Brief from your memory on demand, and reminders when you ask for them.
One short question a day. Your answer becomes memory once you confirm it.
One bar, wherever you add things. Text, voice, attachment, URL, slash command. AI Twin proposes where it belongs and waits for you to confirm.
Bring documents and photos to your review queue. You confirm what becomes memory.
Forward anything to your own inbox address.
Bring your exported conversations into your review queue. You choose what to keep.
A typed entry surface for any entity. Fields are AI-prefilled. You confirm or correct.
Cards keep the records you look for most: policies, documents, things you own. Each one is live from memory, not a copy.
A five- or fourteen-question style assessment at onboarding that tunes AI Twin to write back in your register.
Every meaningful action is logged in an append-only audit log you can read.
Every entity tagged LOW, MEDIUM, HIGH, or CRITICAL. Access rules enforced at the database level via row-level security.
Every AI call runs under a cost cap, with safe degradation.
Added from the workspace switcher once your own Twin is set up.
Personal Workspace and Practice Workspace, with clean scope switching. People you share with see only what you allow.
Start from your company number. The Companies House record comes in, and you confirm each field.
Use ← → to switch features, Home/End to jump to first or last.
One bar, everything
One bar, wherever you add things. Text, voice, attachment, URL, slash command. AI Twin proposes where it belongs and waits for you to confirm.
Universal Input, static diagram
One bar accepts five kinds of input: a question, a forwarded email, a voice memo, a slash command, and a URL. AI Twin classifies each and routes it to the right destination: Person, Event, or Brief.
WHERE IT SITS
Many AI products keep what they learn about you inside their own product. AI Twin sits one layer above the models and holds the memory, so you can change the model without losing it.
Your memory. Typed, audited, yours. It stays here.
Claude, ChatGPT with developer mode, or another assistant that supports MCP connectors. Swap any time.
You bring what matters.
The model is a commodity. The memory is the point, and it is yours.
New to the idea? Start with our guide to AI memory.
Get started
It takes three short steps.
Running a practice? Set up your own Twin first, then add the practice from the workspace switcher. It sits alongside, kept apart.
Everything is free for the first 7 days. No card needed. Add one when you decide to stay.