There’s a concept in economics called the barrier to entry. The thing that stops you from participating — cost, time, skill, access, credential. For most of human history those barriers were largely fixed. You couldn’t be a filmmaker without a crew. You couldn’t architect without years of draughting. You couldn’t run a software company without a team.

AI is taking a sledgehammer to all of that. Not selectively. Not gradually. Everywhere, simultaneously.

What the Barrier Was Actually Doing

Barriers weren’t just obstacles. They were rough proxies for competence. If you’d spent ten years mastering a craft, you probably knew what you were doing. The difficulty was, in part, the guarantee. Getting past it meant something.

That contract is dissolving. Faster than most people are comfortable admitting.

What AI Actually Changes

The lazy narrative is that AI automates repetitive work. That’s true. It’s also the least interesting part of the story.

What AI actually does is compress the distance between intention and execution.

Not automate. Compress. There’s a difference.

Automation replaces a known, repeatable process. What’s happening now is different — AI extends human capability into territory that was previously gated by expertise. A solo developer reasoning across an entire codebase. A one-person charity producing grant proposals and communications at a quality that used to require a department. A researcher synthesising a field they only half-know.

The right question isn’t “what jobs does AI replace?” It’s: what could you never attempt before that you can attempt now?

The Solo Founder Problem, Solved

Take the solo founder. No capital. No team. Historically forced to choose: technical depth or business breadth. One or the other. The rest you outsource or go without.

That trade-off hasn’t disappeared. It’s compressed. The solo founder now drafts at junior-designer quality. Writes copy, investor updates, and legal summaries without outsourcing. Debugs across unfamiliar stacks without becoming fluent in each. Researches markets without hiring analysts. The output isn’t a full specialist team. But it’s good enough, which is exactly what a barrier to entry is about.

Good enough to ship. Good enough to compete. Good enough to be taken seriously.

Domains That Were Previously Closed

Software — You no longer need fluency in a stack to build on it. You need enough understanding to reason about what you’re building and evaluate what comes back. That’s a far lower bar than mastery, and it’s sufficient for most problems.

Legal and regulatory — First-pass contract review, compliance mapping, regulatory summaries. Not a replacement for a solicitor. A replacement for the paralysing ignorance that stops most small operators from engaging with legal questions at all.

Creative production — Video, audio, image, writing. The constraint has shifted from technical execution to creative direction. That’s not a small thing. Those are very different skills.

Research and synthesis — Keeping up with a field, understanding adjacent disciplines, finding the gaps. AI doesn’t replace the scientist. It radically lowers the cost of being a well-informed one.

Data — Turning raw numbers into insight used to require statistical fluency or the patience to develop it. Asking the right question is now worth more than knowing the right tool.

The New Barrier

Here’s where it gets uncomfortable.

When anyone can produce polished copy, polish isn’t the differentiator. Voice, perspective, and genuine insight are. When any developer can scaffold a working app in an afternoon, the code isn’t the differentiator. Product sense is.

The barriers haven’t gone away. They’ve moved upstream — toward judgment, taste, domain knowledge, and the ability to ask better questions.

And that transition is not flat. People who already had depth are the biggest immediate beneficiaries. The experienced architect using AI builds faster. The experienced developer ships more reliably. The new entrant using AI without grounding still produces shallow work — it just looks less shallow than it used to. The levelling is real. But a lever amplifies in proportion to the arm that holds it.

More of Everything

The claim in the title deserves scrutiny. Does AI let us do more of everything, or just the same things faster?

I think it’s the former, and the distinction matters.

When the time cost of a task drops dramatically, you don’t just do the same task faster. You do tasks you wouldn’t have attempted at all. The post you thought about but never wrote. The prototype you sketched but never built. The research direction you noted but never followed.

That’s what “more of everything” actually means. Not a multiplier on the same fixed set of activities. An expansion of the set of things you attempt — and therefore, eventually, of the things you achieve.

Latent intentions become executable.

The Editorial Burden

Lowered barriers carry weight.

When entry was hard, the difficulty did some of the quality control. Now that it’s easy to produce and distribute, the individual carries more of the editorial responsibility. More content. More code. More products. Most of it mediocre. The tools don’t guarantee quality; they remove the excuse of friction.

The people who thrive here are not those who use AI to produce more of the same. They’re those who use it to attempt things worth attempting — and who bring enough judgment to know the difference.

That’s the real barrier now.

It always was.

The tools have changed. The standards haven’t.