Not the company's profile of you — your own, and yours to keep.
We're at the dawn of everyone having a personal AI assistant — you may already have one in Claude or ChatGPT, quietly learning you. That knowledge can't be manufactured: it accrues over years, tracking your moods, your shifting interests, the messy overlap of work and private life. Yet today it's trapped — none of those assistants hands you a portable copy of the you it has learned, only raw chat logs, so switching providers means starting from zero. orqo steps up the other way: every person gets their own Chief of Staff — in a company, each member, walled from the next — and the personality map it builds is private to you, and never the company's to own. Making that map portable too — exportable, carried between tools, locked to no one, not even us — is the road we're on, not a button we ship today.
The Chief of Staff draws on two memories at once — and a hard wall runs between them.
What the whole company knows — its documents, its rulings, its hard-won patterns — classified and connected, shared across the org so any workflow can draw on it. The knowledge graph
The people, groups, preferences, and events around you, and how they connect — held privately, keyed to you alone. The company can't read it; other people can't read it; it's read only when the assistant is talking to you.
The assistant sees both at once — answering from the company's knowledge while remembering who you are — but it never spills one into the other. A fact about the org's world becomes shared knowledge; a fact about your world stays personal. Personal stays personal.
The company's memory is built on a didactic ontology — knowledge sorted by what it teaches and how facts explain one another. Your world isn't a subject to be taught. It's people, relationships, preferences, and the things that happen.
So the personal graph uses its own ontology — but it's the same idea as the company's memory: a typed graph, where every node and every edge carries a meaning the assistant can reason over and traverse, not a flat list of notes. Only the types differ. The knowledge graph classifies knowledge by what it teaches; the personal graph maps the people, groups, and events in your life by how they actually relate — a social vocabulary instead of a didactic one. We call it POER — a Personal & Organizational Entity-Relation graph — deliberately separate from the org's knowledge graph: different types, different privacy model, different owner.
A new chat starts from zero every time. An assistant meant to work with you for years can't.
Because the personal graph persists, the assistant compounds what it learns about you the way a good colleague would — your standing preferences, the people you work with, the history behind a decision. And it doesn't stop at the edge of the office: work and private life overlap, and how you like to work is a personal matter, not a fact about any one project. The preference that you want drafts blunt and short, that Maria handles the EU accounts, that last spring's vendor left a bad taste — those belong to you, and they travel with you from one workflow, one project, one year to the next.
Your graph is divided into rooms — Work, Family, a sports team, a specific project. You never share the whole graph; you invite someone into a single room, they accept, and it becomes a circle.
You and a teammate share a work circle around the account you both own. Each of you contributes your own room, read-only; each of you can read the other's. The assistant serves you both from the union — without merging your private notes into a free-for-all.
Co-parents share a family circle around their child. Everything either of them tells the assistant — what was tried, what worked, what the school said — enriches the shared view, so the two of them pull on one string together instead of briefing the assistant twice.
The mechanics make it safe. Sharing starts with an invite and takes hold only once the other person accepts — never assumed from a chat. It's room-scoped — your other rooms stay private. It's read-only and mutual — a shared view, never a shared owner; John reads Susy's room and Susy reads John's, but neither ever writes into the other's. Consent is explicit and recorded on a consent page. And it's revocable — leave the circle and your view of the others ends, on the spot.
This is the quiet beauty of two graphs with two owners. When someone leaves, the shared knowledge stays — the documents, the decisions, the workflows live in the org's graph, because they're the company's. But the personal graph isn't the company's to keep: it's keyed to you, private to you, and it was always yours. The factual know-how stays with the company; the assistant that knows you goes with you. Knowledge that stays — and knowledge that's yours
Private by default, shared only by consent, and yours to keep. Give one person a Chief of Staff that remembers — or a whole company a way to work together without anyone giving themselves up. Claim your slot