Why I Built
RAGMI.

Nine years ago I started thinking about a problem that most people don't realise they have. Not a technical problem. A human one. What happens to a life when it's over?

I don't mean the legal things, the will, the estate, the arrangements. I mean the texture of a person. The stories they told at the dinner table. The music that meant something to them at twenty-two. The people they loved and lost and never quite got over. The version of themselves they were before you knew them. All of that, the irreplaceable, unrecoverable substance of a human life - quietly disappears.

That bothered me. So I started building.

For the first several years, the work was about capture. How do you take a life, the photos, the documents, the memories, the connections, the soundtrack, and give it structure? How do you make it navigable? How do you build something that says: this person was here, and this is what their world looked like?

I went through many iterations. Applications came and went. The concept, what I came to call a folio evolved slowly. A folio isn't a scrapbook or a photo album or a family tree. It's a container for the full dimensionality of a person. Their timeline, broken into tracks: life events alongside career, hobbies alongside music, the personal threaded through the professional. Their connections mapped and described. Their stories written in their own words, weighted by what mattered most to them.

By the time I had it working the way I wanted, the capture problem felt largely solved. You could build a folio and it would hold a life, faithfully, richly, in a way that felt true to the person it described.

A life story that sits in an archive is just a record. Valuable, yes. Irreplaceable, certainly. But passive. Static. The people who come after you, your children, your grandchildren, people who never got to meet you can browse it, but they can't ask it anything. They can't have a conversation with it. They can't say:

tell me about the year you spent in Canada - or - what did you think of your father - or - what was the song that was playing when you met Mum?

I had nine years of captured lives and no way to make them speak.

Then OpenClaw changed everything.

It just clicked. I was thinking about what an AI agent actually is - not in the narrow, functional sense of a tool that executes tasks, but in the broader sense of a presence. A persistent, intelligent entity that knows things, remembers things, has a perspective shaped by everything it's been given.

After you die, your agent doesn't die with you. It can be maintained. Passed on. It can carry the accumulated context of your life, not as a static archive but as something that can be questioned, that can respond, that can say I remember in a way that means something.

The question then became purely technical: how do you make a life story consumable by an agent? How do you translate a folio - with all its complexity, its significance weightings, its relationships and timelines and narratives - into something an AI can actually use?

Human RAG is the answer.

Retrieval Augmented Generation is how you give an AI access to knowledge that isn't baked into its training. You structure the information, you store it in a way the model can retrieve, and suddenly the agent knows things it didn't know before. RAGMI applies that same principle to a person. Your folio becomes a structured context layer - a Human RAG - that your agent can draw on in real time. It knows who your oldest friends are. It knows what your favourite football team means to you. It knows the arc of your career and the music of your twenties and the way you thought about the world.

While you're alive, that means an agent that's genuinely useful rather than generically capable. An agent that feels, over time, less like a tool and more like a companion.

After you're gone, it means something more.