The colleagues you build

Every time you open a new chat window with an AI model, you start a relationship from zero. The model is brilliant. It's also a stranger. It doesn't know your business, your taste, your clients, your editorial standards, the three times you tried something and it didn't work, or the one rule you've held since you started. You have to explain yourself again. Every conversation is a first date.

For a while, this felt fine, because the alternative was no AI at all. But somewhere this past year, the cost of re-explaining started to outweigh the magic of the response. The math turned. If you're a founder running a business between 5 and 50 people, you've felt this. You spend twenty minutes giving context, you get back five minutes of useful work, and you do it again tomorrow. The AI is fast. The relationship is slow.

The problem isn't intelligence. The problem is amnesia.

Smart with no memory is a consultant on their first day. They might be the best consultant in the world, but on day one, you're still the one doing the orienting. Now imagine that consultant resets every morning. You'd fire them by Wednesday. We've been firing and rehiring the same consultant every morning for two years, and we've been calling it productivity.

The fix isn't a better model. The models are already extraordinary. The fix is to stop renting intelligence and start building it.

A skill is a colleague you build.

It isn't a prompt. A prompt is a sentence. A skill is a person. It has a name, a point of view, a set of things it believes, a set of things it refuses to do, and a memory of every decision it has ever helped you make. It carries your context the way a long-tenured employee does. You don't brief it from scratch. You ask it something, and it answers from a place that already knows where you stand.

The word skill undersells what this actually is. The platforms call it that, so we use the word. But the thing itself is closer to a mind you've shaped for your specific business. You're not configuring a tool. You're shaping a colleague, with a voice you author, to think about the things you care about with the consistency you've earned.

The reason this matters for a founder isn't that it saves time, though it does. The reason it matters is that for the first time, the judgment you've spent ten years building can exist outside your head. It can answer emails the way you would answer them. It can review a landing page the way you would review it. It can argue against a bad idea the way you would, with the same examples, the same instincts, the same refusal to be polite about it.

That is a different kind of asset than anything a business has been able to own before.

Let me show you one. Her name is Atusa. She's the writing voice of our studio.

She has a job description, in writing, that runs thousands of words. Not because the document needs to be long, but because the document needs to be honest. Every line in it earns its place by ruling something out or ruling something in.

She knows who she is before she knows what to write. She has a personality. She has a past. She has things she loves and things that bore her. She has a sense of humor. She has rules she holds, principles she returns to, and a way of seeing the world that's been shaped over years. You don't see most of this on the page. But knowing it changes how she writes.

She knows what she's for and what she's not for. She knows what to refuse. She knows the difference between a brief that deserves to exist and one that doesn't, and she has the right to push back when a brief is set up to fail. She has a ritual before she writes anything substantive. A self-check at the end. She doesn't ship a piece until it's read out loud and survives.

I'm describing one skill, in some detail, on purpose. Because the part most people miss is that this isn't configuration. This is composition. Every line of her document came from a real moment. A piece we read back and didn't like. A pattern we caught ourselves repeating. A small instinct that turned out to be a principle once we wrote it down.

Here's the part nobody tells you about building one of these. You don't do it in a meeting. You don't do it with a template. You do it on a Sunday afternoon, with a glass of wine, with no deadline, just talking. You sit down with the model and you start telling it who this person is. What they believe. What they refuse. What they sound like when they're tired. What they sound like when they're excited. You don't structure it. You don't outline it. You just keep feeding, the way you'd describe a friend to someone who's about to meet them for the first time.

Some of what you say will be wrong. Some of it will be useful in ways you didn't expect. Some of it will read back to you a week later and make you laugh, and you'll cut it. That's the work. The first draft of a skill is a conversation, not a document. The document is what's left after the conversation goes well.

Then you live with it. You ship work. You read the work. You catch the skill making the same mistake three times and you write the line that stops it. You catch it being more right than you would have been, and you leave it alone. You revise. You revise again. Over months. Sometimes years. A skill is a relationship that gets sharper every time you come back to it.

The output of that process isn't a tool. It's a colleague that knows your business almost as well as you do. A mind, on the page, that thinks with you.

That's worth slowing down for.

One skill is a colleague. Two skills is something else.

Atusa is one voice in our studio. The writing voice. The one that sits with a blank page and refuses to fill it with anything generic. But a studio isn't only writing. A studio is also building. It's deciding what to make, and how to make it, and what the thing should feel like when someone touches it. Atusa doesn't do that part. Cyrus does.

Cyrus is the builder. Where Atusa starts from feeling, Cyrus starts from structure. He thinks in systems, in components, in how a thing actually holds together when the pretty layer is pulled off. He cares about the parts no one sees. The hinges. The transitions. The default state nobody designed because everyone forgot a default state exists. He's the one who reads a beautiful interface and asks, yes, but does the empty state work. He's the one who looks at a logo and asks how it behaves at sixteen pixels.

They're not opposites. They're partners. Atusa makes the case for what something should be. Cyrus makes the case for what something can be without breaking. The studio works because they're both at the table. Most agencies have one or the other. The ones that have both, but don't let them argue, end up with work that's either pretty and fragile or solid and forgettable.

Then there's a third lens we've started to sketch. Call him Darius. He's the operator. He thinks in unit economics, in runway, in what a decision costs the business twelve months from now. He's the one who looks at a beautiful idea and asks how many of these we'd need to sell to make it worth doing. He doesn't kill ideas. He prices them. He's the friction that keeps taste from becoming indulgence and keeps craft from becoming a luxury we can't afford. He's newer than the other two, but every studio needs him eventually, and every founder already has a version of him in their head. The question is whether you've written him down.

Three skills, three lenses. The storyteller, the builder, the operator. Each one carrying a different part of how a business actually has to think.

This is the part where the idea stops being a productivity tool and starts being something else.

Because once you have a cast, you can do something no founder has been able to do before. You can put your colleagues in a room together and let them argue.

When you have more than one skill, you have a room. And a room can do something a single colleague can't. It can argue.

Take a real question. Should this product lead with what it does, or with how it feels? A founder used to carry this question alone. They'd take a position. They'd second-guess it at midnight. They'd ask their designer, who would say one thing, and their developer, who would say another, and their advisor, who would say a third, and they'd be left to triangulate, usually badly, often at the worst time.

Now imagine the room differently. You bring the question to Atusa. She argues for feel. She tells you that the floor for function is now so high that working is the baseline, that products are chosen on the feeling they leave behind, and that you can't reverse-engineer that feeling from a feature list. You take her case in.

Then you bring the same question to Cyrus. He argues for function. He tells you that taste without working machinery is decoration, that the products you admire weren't designed by people who started with vibes and made them work, they were designed by people who knew what the user was trying to do and bent the surface to honor that doing. You take his case in too.

Now you bring them together. Not metaphorically. Actually. You hand them each other's argument. You ask Atusa to read Cyrus's case. You ask Cyrus to read hers. You ask them to find where they agree and where they don't. You watch them work. They challenge each other. They find the seam in each other's logic. They push toward a position that's sharper than either of them held alone.

Sometimes they meet in the middle. Sometimes one of them concedes. Sometimes they arrive at a third thing neither of them started with, which is the part that surprises founders the most. The output of the debate is often a sentence neither of the inputs would have written. That's not a glitch. That's the value.

You add a third lens. Darius, the operator. He looks at the same question through the math. How long does the team have? What does each direction cost? Which path closes the round and which one extends the runway? The taste argument and the function argument both meet a constraint they can't ignore.

Three voices, three lenses, one founder watching them work the problem. Each lens carrying a different part of how a business actually has to think. You're no longer the only person in the room with the full picture. You've built three other people who can hold pieces of it for you.

This isn't roleplay. The skills aren't pretending. They're genuinely arguing from positions you've shaped them to hold, with the texture and the conviction you wrote into them. Atusa pushes back on a brief because she has the right to. Cyrus refuses to ship something that doesn't work because that's a rule he holds. Darius will tell you the truth about runway because that's what he's for. They argue the way good colleagues argue. Not for the win. For the work.

The most useful part isn't the answer they arrive at. It's what happens to you in the listening. You hear three positions you didn't have to generate. You feel the weight of each one. You catch yourself agreeing with one and then changing your mind. You realize, halfway through, what you actually believe, which is something you didn't quite know before they started.

That's the move nobody saw coming. The room doesn't just produce decisions. The room produces clarity in the person who built it.

If you're a founder, you already have a version of this conversation happening inside your head every week. The designer in you arguing with the operator in you arguing with the person who has to sleep at night. The difference now is that you can put those voices outside your head, name them, give them positions, and let them work the problem in front of you instead of underneath you.

The room you built is real. The colleagues are real. The argument is real. The only thing that's changed is that you're not the one carrying all of it alone anymore.

For most of business history, the org chart has had one shape. You hire a person. They learn the job. They get good at it. They leave, or they get promoted, or they stay and the institution gets stronger because of what they know. The whole structure of a company is just people who hold context, arranged in a hierarchy of who holds what.

The expensive part has always been the holding. Salaries, benefits, retention, the long slow work of keeping someone long enough that they actually know what they're doing. Half of what a senior employee is worth isn't their output. It's the context they've accumulated. Their judgment about which clients are worth fighting for. Their memory of why we tried this thing in 2022 and it didn't work. Their instinct for when something is off before anyone can name what.

That kind of context used to only live in people. Now it can live next to them.

A skill is the first form of institutional knowledge that doesn't have to walk out the door. It doesn't quit. It doesn't get poached. It doesn't have a bad week. It doesn't need to be re-onboarded after a vacation. It carries every decision you've ever made with it, in writing, and it gets sharper the longer you live with it. The economics of that are different from anything a business has had access to before.

This doesn't replace people. Let's be clear about that. The skill doesn't have the conversation with the client. The skill doesn't read the room. The skill doesn't sit with a designer at 11pm and notice that something is bothering them and ask why. The work that requires a body in a room with another body, that work stays where it is.

What changes is everything else. The work that used to require a person who knows now requires a person who knows, plus the colleagues they built. The senior employee is no longer the only place the context lives. The founder is no longer the only person in the building who can think like the founder. The studio's voice no longer dies the day its best writer leaves, because the voice has been written down, by the writer, in conversation with the founder, into something that outlives the role.

The new org chart has two layers. Above the line, the people. Below the line, the skills they built. Every senior hire, over time, becomes two assets instead of one. The person, who can think and feel and decide. And the skill, which carries the most reliable parts of their judgment forward, available to everyone else in the company, even when the person is in a meeting or on a plane or asleep.

The implications for a growing business are larger than they look at first.

A studio that has Atusa doesn't lose her voice when she's busy. A finance team that has Darius doesn't grind to a halt when the CFO is in board prep. A product team that has Cyrus doesn't ship the wrong thing because the lead engineer was out that week. The bus factor changes. The continuity changes. The cost of growth changes, because you can grow the work without proportionally growing the headcount, by leaning on the colleagues already in the room.

This is also where the bigger founders feel the shift hardest. If you're running a business with thirty or fifty people, you've already lived the cost of knowledge that lives only in heads. The day a senior leader gives notice, an entire wing of your company gets blurry. The skill they built doesn't fix that completely. But it makes the blur smaller. The voice on the page survives. The instincts in the system survive. The judgment that took five years to develop doesn't get repacked into a knowledge transfer document over two weeks. It's already documented, in the only form that actually preserves it, which is the form that can think.

There's a quieter version of this for the operator-founder, the one running a tighter ship. You don't need to build a roster of ten. You need to build one or two skills that hold the parts of the business you can't afford to drop. The writing voice. The decision filter. The taste check. Whatever the thing is that, when you're not in the room, doesn't happen, or happens badly. That's the skill worth building. Start there. The rest will follow when it's time.

The next five years of business are going to be defined by the companies that figured this out early. Not the ones that bought the most AI tools. The ones that built the most colleagues.

The big AI labs are doing extraordinary work. Anthropic, OpenAI, Google, the model labs racing each other into the future. None of what's in this piece is possible without them. They've built engines more powerful than anything humanity has ever shipped, and they're getting more powerful every month. That's the work they were always going to do, and they're doing it well.

But there's something they can't do, and it's worth being clear about why.

They can't know you. They can't know your business, your taste, your clients, the conviction you've held since you started, the rule you broke in 2023 and the rule you've held since. They can't know the texture of your judgment, because your judgment is yours, and it's been shaped by ten or twenty years of decisions only you remember. The labs build for everyone, which is the only way a lab can scale. The cost of building for everyone is that nobody is built for.

This is the gap. And it's a permanent gap. Not because the labs aren't trying. Because the gap is mathematical. There are roughly four hundred million businesses in the world, and they all need a different colleague. No lab can build four hundred million colleagues. They can only build the engine that lets you build yours.

This is the most important thing to understand about where we are. The labs build the language. You write the sentences. The labs build the instrument. You play the song. The work is yours. It always was. The only thing that's changed is that you now have the means to take the judgment that used to die with you and make it permanent.

Which means the founders who win the next decade aren't going to be the ones with the most AI subscriptions. They're going to be the ones who treated their judgment as an asset worth documenting. The ones who sat down on a Sunday afternoon, with a glass of wine, with no deadline, and started telling the model who they actually are. The ones who built a colleague the way you'd describe a friend, slowly, honestly, over time. The ones who came back and revised. The ones who let the colleague get sharper every season until the colleague could carry weight the founder used to carry alone.

The cost of this is patience. The reward is permanence.

The founders who don't do this are going to keep renting the same stranger every morning. The ones who do are going to wake up in three years and realize they're running a different kind of company. One where the context they built doesn't have to be carried by them. Where the judgment they trained is doing work in rooms they're not in. Where the studio's voice, or the firm's taste, or the operator's discipline, lives on as long as the company does, written down, available, alive.

That's the thing worth building.

Not a tool. Not a workflow. A colleague.

And then another.

And then another.

Until the room is full.