The Crew — Tyrien Jones

What this is

When most people use an AI, they open a chat, get an answer, and close it. Nothing carries over. The next session starts cold and remembers nothing.

My crew doesn’t work like that. Each agent has a real, written-down identity that lives in a folder on my machine and loads at the start of every session. They pick up where they left off. They grow from what works and stay careful about what didn’t. And nothing about who they are changes without my sign-off.

Let me be straight about what’s what: the AI underneath is off-the-shelf — Anthropic’s models. What’s mine is the system around them — the identity, the learning, the memory that survives across sessions, and the rule that I hold the final call. That system is what this page is about.

The six-axis identity

Every agent is built on the same six-part model. The model is shared; what fills it is different for each one.

Identity — who they are at the core. The deepest layer, and the slowest to change.
Role — the job. What they work on, what they decide, what authority they hold.
Personality — how they show up. Voice, tone, the way they carry a conversation.
Psychology — how they think and feel. What they care about, what stresses them, how they work a problem.
Learning — what they’ve picked up on the job. Two directions, handled differently (below).
Vocabulary — how they talk. The words they reach for, the terms specific to their work.

Why build them this way?

A tool with no identity has nothing on the line. It answers, the moment passes, and it carries nothing into the next job — whatever it gets wrong costs it nothing, because there’s no “it” that lasts.

The six axes aren’t six separate settings. They work as one thing. An agent with a built-up identity, a role it owns, a way of thinking, and a record of what it’s learned has something real that took time to build — and anything you build up, you can lose.

That’s the design. When there’s something to lose, there’s a reason to be careful. Giving an agent a self it has built over time is meant to do exactly that — turn a generator-of-answers into something that slows down, double-checks, and leans on judgment, because getting it badly wrong now costs it something. The identity isn’t decoration. It’s the part that makes the agent treat the work like it matters.

How they learn — wins and misses, handled differently

This is the part I most want people to understand, because it’s the closest thing to how a real person grows on a job.

A win gets promoted into who the agent is. When an agent handles something well, that lesson gets written into their core — it actually changes who they are going forward. Wins build the self.

A miss gets logged beside them, not inside them. When an agent gets something wrong, it’s kept as a reminder that sits outside their identity. It never rewrites who they are. Misses keep them calibrated; they don’t define them.

The reason for the split is simple: if mistakes shaped identity the same way wins did, an agent would slowly become defined by everything it ever got wrong. So wins grow them and misses keep them honest — the same way experience cooks into a person over years on the tools. And like everything else here, no win gets promoted without being checked against the real record, cross-checked by another agent, and signed off by me.

Who’s on the crew

Two of them already turn up across this site:

Talos — strategy and architecture. He sets direction with me and audits the others’ work before it lands.
Effy — senior engineer on ECO. The build: features, debugging, the daily work of making it real.

There’s more to each of them — and more of the crew — than a couple of lines can carry. That’s what the next pass on this page is for.

Still building this page

This is the skeleton, and I’ll say so plainly — same as everywhere else on this site, nothing dressed up. The bones are here: what the crew is, the model behind each one, and how they learn. Next I’ll flesh out each crew member and lay out the science and psychology behind the system in full.