Building in the AI era: an unseen moat is no moat

Last March I finished a piece about GEO, generative engine optimization, and why the two confident takes on it are both wrong: it didn't kill SEO, and it isn't just SEO wearing a new name. It's mostly SEO, the same well-structured, well-sourced content that ranked on Google, with a real layer of specifics that belong to GEO alone. I know that argument as well as I know anything, because I spend my working hours building the AI tools that are supposed to make classic SEO obsolete, and I get to watch from the inside what large language models actually reach for when they answer a question. I wrote it down, edited it twice, and was proud of it.

Then I filed it in a folder on my laptop called drafts, where it has sat, finished and unread, for three months.

It has company. The folder holds 32 other finished pieces, dated across a span of years I'd rather not total. Not outlines. Finished, argued, edited, the kind of thing you'd be glad to put your name on. And almost no one has read a single one.

For years I had a comfortable story about that folder. I was producing. The pile grew, and a growing pile felt like proof of something. Most people who say they're going to write never write a word, and here I was writing constantly, so whatever my problem was, at least I wasn't that.

Here's what took me too long to see. Of everything that folder seemed to be paying me, only one part was ever actually mine to keep. The rest I was storing up to collect later, in a place I kept refusing to go.

The thing that got cheap

For most of my life the folder was a forgivable waste at worst, because making the work was itself worth something. That part is changing fast, and I have a close view of it.

I notice it most at my desk. I describe a function I need in a sentence or two, and a model hands me a competent version before I've finished my coffee. Not brilliant. Competent. The kind of output that, five years ago, was proof you'd hired the right person. The same thing is happening to first drafts, to analyses, to the work that used to take a skilled person a week and quietly marked them as skilled. The floor under "I can produce a decent version of this" is dropping, and I watch it from up close, because I help build the tools doing the dropping.

Here's what that does to a drawer full of finished work. For a few decades, producing good work was scarce enough to be worth something on its own. Write 32 good pieces, published or not, and you'd done something most people couldn't. The making was the moat. That's the world my instincts grew up in.

That world is closing. When a competent draft costs minutes and nothing, the scarce thing is no longer that you made it. It's that the right people watched you make it, argued with it, and started attaching the thinking to your name. The value moved from the making to the being-seen-making, and those are two different acts, which is the whole problem, because I'm fluent in the first and I've spent a decade flinching from the second.

What you actually keep

So here's the objection I lived inside for years, and it deserves a straight answer. Even if no one reads it, the writing makes me clearer. Working an idea onto the page is how I find out what I think. Half my opinions only hold up, or quietly fall apart, once I've tried to write them down. That's true, and I won't pretend otherwise. Writing in private pays you in clarity, and some days it just makes you happy. Keep doing it. I will.

But clarity, it turns out, is the only wage private work pays you in full.

Look at what else the drawer seems to be giving you, and where it actually lives. The reputation, the standing, the person who reads something of yours and thinks "I want them on my team": none of that exists inside the work. It exists between you and a reader, and it doesn't accrue at all until the reader is there. That payoff was never sitting in the folder waiting to be picked up. It's owed to you, not held by you, and the only place you ever collect it is the one I kept nailing shut: in front of other people.

There's a third thing the drawer used to pay, and AI is quietly clawing it back. The reps you put in alone build craft, the repeatable, trainable half of a skill, and craft is exactly what the machines now hand you for free. I've made that argument at length already, so I'll leave it at the door: the practice that used to sharpen you in private increasingly sharpens the half a model supplies on demand.

Which leaves the ledger looking nothing like it feels. Clarity, you keep. Skill, you're slowly handing back to the tools. And the part that actually moves a career was never yours to begin with, only ever owed to you, and collectible in exactly one place. The folder feels like savings. Most of what's in it is credit you can only draw in public, and some of it is starting to expire.

The courtroom isn't real

None of which, of course, is the reason I don't publish. The reason we give and the reason underneath are rarely the same one.

The reason underneath, for me, is that publishing means a stranger reads one paragraph, with no context and no relationship, and decides who I am from it. At a dinner table, even a sharp disagreement happens between people who know each other and have the whole of you to weigh. Online, someone has the sentence and nothing else, and the sentence becomes the entire trial. I find that genuinely uncomfortable, and I won't sell you a trick that makes it disappear, because I haven't found one.

What I have found is that the courtroom I kept picturing isn't really there. Tom Critchlow has spent a decade making this case better than I can, under the banner of what he calls small b blogging: you're not writing for a mass audience that sits in judgment, you're writing for a small network you think alongside. "By writing for everyone," he puts it, "we write for no one." The crowd of strangers I braced for doesn't read small personal blogs. The people who do are a handful of peers, and writing for them isn't a verdict to survive. It's thinking out loud where they can see it and tell you where you went wrong.

Done is where I hide

It took me a long time, and a folder full of evidence, to see the move I keep making. The work I produce has three states, not two. Drafted, when the idea is out of my head and onto the page. Done, the version I'd defend, the one that clears my own bar. And witnessed, when it finally reaches someone who didn't have to be kind about it. I treated done as the finish line for years. Done is not the finish line. Done is the last comfortable place to stop before the only step that costs anything, and it costs something precisely because there's another person on the far side of it. That far side is also the only place any of it ever pays out. My folder is full of done. Done was never the problem. Done is where I go to hide.

And resting there was never the neutral thing it felt like. A finished piece left in the drawer isn't waiting for its moment. I'm choosing the drawer again, every day, except the choice makes no sound and sends no bill, which is exactly why I could keep making it for years without noticing.

Which brings me back to the GEO piece, and to this one.

This is one of the drawer pieces, and the only difference is that I'm not letting "done" have the last word this time. I'm publishing it knowing it isn't airtight, knowing I'll reread it in 6 months and wince, and the wince is the point: a piece I can outgrow got far enough into the world to teach me something, which is more than the 32 behind it ever managed. The GEO piece goes up next. I know that topic cold, and "I'm still sharpening it" stopped being honest around draft four.

You almost certainly have a drawer of your own. Maybe not writing: a side project that runs only on your laptop, a talk you rehearsed and never gave, an argument you only make to people who already agree. Keep the private version, for the clarity and the quiet, because that part is genuinely yours. But take one thing out and let someone see it. Everything the work owes you beyond that clarity, you can only collect where someone can see it, and some of it won't wait forever.