When everyone can automate, what’s left that’s yours?

I build AI systems for a living, and I’ll be honest: some of them are designed to take a person out of the loop. At work I’ve watched analyst roles shrink. A company that would have hired a team of data analysts a few years ago now hires a handful and asks a model to do the rest. I’m one of the people building that model.

So a question follows me around. When everyone can automate — when a model can write a paper, a dashboard, or a portfolio project in minutes — what is actually left that’s yours?

The pattern I keep noticing

Ask ten early-career students for their résumés and they look almost the same: one retrieval project, one optimization project, one model fine-tuned with the tutorial everyone followed. It’s not that these people aren’t capable. It’s that nobody ever told them what was actually worth building. So they copy the tutorial, never reach the real problem underneath it, and skip the fundamentals that would have set them apart.

You see it everywhere once you start looking. The student told to “just build projects” who has no idea which ones. The endless stream of papers and news with no way to tell what actually matters for your goal. Everyone knows GPUs are important, but few get to the depth where it counts; everyone is hiring for LLMs and agents, and most people never get past the surface of either.

And this isn’t only a problem for engineers. It’s the same for the arts student, the person switching careers, the one who decides to “learn Python now” without any real sense of what advantage that’s meant to build. I’m not thinking about the small group who’ve already figured it out — they’ll be fine. I’m thinking about everyone else: the ninety-five percent who are looking for an edge and can’t find one.

It isn’t a memory problem

When I describe this, people often hear “memory product” — something that remembers your preferences and gives you more of what you already like. I think that’s the wrong instinct, and a slightly dangerous one. A system that just feeds you more of what you’ve already chosen doesn’t help you grow. It quietly smooths you into a more comfortable version of average.

Remembering you isn’t the hard part. The hard part is helping you become someone.

What’s missing is agency — your autonomy, your originality, the thing only you would make. I’ve come to think of agency as the variable that sits upstream of everything else. Your work, your sense of meaning, your creativity, what you learn — all of it flows from how much real authorship you have over your own direction. Get the upstream right and the rest tends to follow.

This isn’t only my hunch. When Anthropic interviewed around eighty thousand people across the world, the two things they most feared losing to AI — after plain unreliability — were their agency and their ability to think for themselves. People don’t want to be replaced. They want help becoming more.

How agency actually shows up

Agency isn’t abstract. Other people recognize it through two things: originality and craft. Originality is the what — a question only you would ask. Craft is the how — the depth and care you bring to answering it. A clever idea executed sloppily doesn’t land, and a flawless copy of someone else’s idea just makes people shrug. Only the combination reads as real.

This is the part AI changed. It made the output cheap. Anyone can mint a paper or an analysis now, so the finished artifact no longer proves much about you. What AI can’t hand you is the rigor and the original question behind the work.

The goal isn’t to use AI to produce. It’s to use AI to become.

Automation isn’t the villain here, either. If a tool clears the boring eighty percent of the work quickly, that’s space — the same way finishing the day’s work frees up the evening. The real question is what you do with the space. Most people never notice they’ve been handed any; they spend the week producing things a machine could have produced, and call it a job.

What coocBoard is

coocBoard is a workbench. The “board” is just the surface; the part that matters is the “cooc” — creating order out of chaos. It’s a personal, local space where you put down your scattered interests and it helps you frame them.

Framing is the whole point. More freedom isn’t what people need — raw AI already offers infinite freedom and leaves you paralyzed. What helps is the right boundary: enough structure that you don’t drown, enough room that you can still be original.

It doesn’t think in tasks. It thinks in arcs of work — a thread of experiments around a question that builds on itself over time. Each arc runs the same simple loop. First it clears the noise, both the world’s and your own, so you can actually see what you’re working with. Then it pushes you to turn that into something real and visible, because agency only counts when it can be seen. And before you call anything done, it asks two questions: so what — does this stand out? — and how rigorous — is it actually deep, or just finished? When something passes both, you ship it. Learn, do, scale; the board grows with you.

Why me, and why now

I believe building is how you become. You don’t think your way to clarity and then start; you build your way there. So I’m making coocBoard as my own workbench first. I’m living the problem right now — caught between molecular dynamics and world models, trying to find the thread that’s genuinely mine rather than the one that happens to be trending. If the tool can help me do that without flattening me toward the hype, then it works.

It’s also, honestly, the whole shape of what I care about: research on AI that serves people, the psychology of how we decide, and the engineering to make it real. That intersection is where I’ve always liked to work. The most interesting problems tend to show up when two worlds start explaining themselves to each other.

An invitation

These are first thoughts — a problem I’m sure is real, and a shape I’m still working out. I’m building it in the open. If any of this feels familiar — the sense that the tools are everywhere but the direction is missing, that producing has gotten easy while standing out has gotten hard — then this is being built for you. The ninety-five percent, not the five.

← Back to coocBoard