Here is something I keep noticing when I work with people outside the AI Twitter bubble.
I will be talking to a smart, successful business owner — someone running a real company with real revenue — and I will ask them how they are using AI. And often the answer is: not much. Maybe they have tried ChatGPT a few times. They have heard a lot about it. But they have not built anything with it. They are not sure where to start.
This surprised me at first. I assumed everyone was further along. When you spend time in tech circles, you are surrounded by people building agents and debating model benchmarks and sharing prompting techniques. It is easy to assume that represents the whole market.
It does not. Not even close.
The Access Problem Is Solved
ChatGPT is free. Claude is a browser tab away. Gemini is built directly into Google products. Perplexity is free for most searches. The tools are available to anyone with an internet connection, at essentially zero cost.
There is no access problem. That gap closed fast.
The gap that exists now — the one that actually matters — is skill.
Skill meaning: knowing how to use these tools to get real results. Knowing how to write prompts that produce useful outputs. Knowing how to connect tools into workflows so things happen automatically. Knowing when to use which model for what task. Knowing how to build an agent that runs without you.
That is a real gap. And it is not closing at the same rate as access.
The Demand Is Real
Earlier this year I started running AI workshops in Austin. I was not sure what to expect — I put together a session, invited some contacts, and sold tickets.
It sold out immediately.
I ran a second one. Sold out again. Five of the attendees from the first workshop came back for the second one, just to stay current on what had changed.
That is not what happens when a topic is saturated. That is what happens when there is a real gap between what people know and what they need to know.
Why the Power Law Matters
Here is the part that does not get enough attention.
The distribution of results from AI is not a bell curve. It is closer to a power law.
A bell curve would mean: people with strong AI skills are maybe 20-30% more productive than people with average AI skills. That is meaningful but not alarming.
What I actually see is more extreme. The people who have really figured out AI — who have moved through the progression from basic prompting to workflows to building autonomous agents — have a highly concentrated amount of leverage compared to people who have not. They are not 20% ahead. They are operating at a different scale entirely.
This makes sense when you think about it. The benefits of AI compound. If you are using AI to generate content, research, and analysis, and you are also using agents to automate repetitive work, and those agents are saving you 239 hours per week — that is not an incremental advantage. That is a structural one.
And structural advantages are hard to close once they are established.
The Three Levels — Where Are You?
When I work with people on this, I use a simple three-level framework to figure out where they are and what the next step looks like.
Level one is AI Assisted. You are using AI tools for individual tasks. You prompt, you get a result, you use it. This is where most people are. It is useful — probably a 1.5-2x multiplier on certain tasks — but it does not compound.
Level two is AI Workflows. You have started connecting tools and building repeatable processes. Instead of one-off prompts, you have systems: a research workflow, a content workflow, a meeting follow-up workflow. This is where things start to get interesting. You are not just saving time on individual tasks — you are removing whole categories of work.
Level three is Building Agents. You are deploying AI that runs without you. You define what needs to happen, set up the system, and come back to finished outputs. This is where the big leverage numbers come from.
Most people are at level one. The gap between level one and level two is where the most accessible productivity gains are. And most people have not crossed it yet.
What to Do About It
The access problem is solved. You do not need to wait for better tools or lower prices. Everything you need to meaningfully move up this progression is available right now.
What is needed is the investment in skill. That means time spent learning, building, and iterating — not just reading about AI but actually using it to build something real.
The window where early adopters have a meaningful advantage does not stay open forever. In most technology curves, early movers get the biggest gains, and then those gains compress as the skill becomes more common.
We are still in the early part of that curve for AI fluency — particularly at levels two and three. The gap is real. The tools are available. The question is whether you are investing in closing it.
