Most People Are Using AI at 10% of What It Can Do

I was talking with an entrepreneur recently — someone who’d built a serious education platform, backed by a team of former Google engineers — about how she teaches people to use AI. What she described wasn’t a course on writing better prompts. It was a school where people build their own working tools within days, graduating only once they’ve built eleven of them with their own hands.

That number stuck with me… Eleven! Not “watch a video about AI”, or “here are ten prompts to try.” Eleven real, working things you built yourself before you’re considered to have graduated the first level. That’s a completely different relationship with the technology than the one most operators currently have, and the gap between those two relationships is going to matter more every year.

What struck me most wasn’t the number itself, it was what the number implied about how she thinks about learning in general. She wasn’t interested in whether people understood the concept of AI in the abstract. She was interested in whether they could point to eleven specific things they’d made. That’s a builder’s standard for competency, not a student’s, and it’s exactly the standard I think more operators need to hold themselves to.

Two Very Different Relationships With the Same Tool

Most people I talk to use AI the same way: they open a chat window, type a question, get an answer, and move on. It’s useful. It saves time. It’s also the shallowest possible use of what’s actually available, and most people don’t realize there’s a next level, because nobody’s shown it to them.

The next level looks like this: instead of asking a general AI tool a question and hoping the answer fits your specific business, you build a tool that’s shaped around your specific business from the start — one that knows your positioning, your customer objections, your internal processes, your team’s specific workflow. The difference between asking a generic assistant for advice and having a tool built specifically around how your business actually operates is enormous, even though both technically involve “using AI.”

Here’s the analogy that made this click for me: everyone who’s ever picked up a guitar knows a few chords. Very few people ever learn to actually build a song, arrange it, produce it, make something that didn’t exist before out of the same six strings everyone else has access to. Most people’s relationship with AI right now is at the “knows a few chords” stage. The operators who are going to pull ahead over the next several years are the ones willing to go further — not because they’re more technical, but because they decided the extra effort was worth it.

I think this distinction gets missed constantly in conversations about AI, because most public discussion treats it as a single monolithic thing — you either “use AI” or you don’t. In reality, there’s a wide spectrum between typing a casual question into a chat window and building a real system tailored to your business, and almost everyone I know is clustered near the shallow end of that spectrum without realizing how much room there is to grow.

What Building Actually Looks Like

I want to be direct about something: building your own AI tools is far more accessible than it sounds, and it doesn’t require a technical background. The barrier isn’t skill. It’s the decision to go past the first, easiest layer of using the technology.

A working AI tool, in practice, is usually a fairly simple structure: clear instructions about what the tool should do, some information specific to your business fed into it, and a way for you or your team to interact with it repeatedly instead of starting from scratch every time. That’s it. The complexity people imagine when they hear “build an AI tool” is almost always higher than the actual complexity involved.

Where people get stuck isn’t the technical part — it’s not knowing what to build first. My advice: start with whatever task you or your team does on repeat, the one that eats real time every single week but never quite justifies hiring someone new to handle it. That’s almost always the first tool worth building, because the return is immediate and the task is well understood enough that you already know exactly what a good outcome looks like.

Once you’ve built one, you’ll notice something interesting: the second one takes a fraction of the time the first one did, because you’re no longer learning the underlying structure — you’re just applying it to a new problem. That compounding effect is the real payoff of going through the uncomfortable first attempt. Most people quit before they ever get to feel it.

The Community Effect

One detail from that conversation stood out more than anything else: the platform isn’t just a curriculum, it’s a community. Everyone’s in it together, at different levels, helping each other work through problems in real time. Someone gets stuck building something, posts it, and within minutes someone further along sends back exactly what they need to get unstuck.

This matters because building anything new is uncomfortable, and most people quit right at the point where a single piece of missing information would have unlocked the whole thing. A community around the learning process solves exactly that problem — it turns “I’m stuck and I’ll figure it out eventually” into “I’m stuck and someone who’s already solved this will help me in the next ten minutes.” That structure accelerates the entire learning curve far more than any course content on its own ever could.

I think this is an underrated part of how any real skill gets built, not just AI. The founders I know who learned something genuinely difficult — a new sales methodology, a new operational system, a completely new industry — almost always did it inside some kind of peer group, not alone with a manual. Isolation is where good intentions go to stall out.

Why This Matters More Than It Looks Like It Does

I think about the operators I’ve watched go through real transitions before — the shift from print to digital marketing, from in-person sales to remote, from analog processes to automated ones. In every one of those shifts, the operators who moved early and built real competency had a window where their advantage over competitors was significant, before the rest of the market caught up.

We’re inside exactly that kind of window right now with AI, and it’s closing faster than the previous ones did. The gap between someone using AI at the surface level and someone who’s built five or six tools specific to their own business is already a meaningful competitive advantage. In another two or three years, it may be the baseline expectation rather than an edge — which means the operators building real competency now are the ones who’ll still be ahead when everyone else catches up to where they started.

Where to Start This Week

Pick one repetitive task in your business. Not the most complex one — the most repetitive one. Spend real time this week learning what it would take to build a simple tool around just that task. You don’t need to master anything broad. You need one working example, built by you, that proves the concept inside your own business.

Once you have one, the second and third get significantly easier, because you’ve already done the hardest part: believing it’s actually within reach. That belief, more than any technical skill, is the real barrier standing between where most operators are right now and where they could be in six months.

We’re bringing builders — real ones, showing real systems — to BB&C 12 this October in Louisville. Day 1 is Revenue, Day 2 is Leadership, and this exact topic is part of the conversation. Register here!

Scott Joseph is the founder of Me Plus Ultra, a private leadership community for entrepreneurs and executives navigating complexity, scale, and high-stakes decision-making. The group works with founders who are no longer looking for motivation or tactics, but for sharper thinking, better judgment, and environments that demand accountabilit

Scott is the author of ASCENT, a book on how leaders rise by changing the rooms they operate in, and why environment, not effort, ultimately determines growth. Through Me Plus Ultra, he curates disciplined conversations where experienced leaders pressure-test assumptions, challenge one another’s thinking, and leave with clarity they can actually execute.

In addition to leading Me Plus Ultra, Scott has founded and scaled multiple companies, built and exited complex operating businesses, and facilitated thousands of executive-level conversations across industries. His work now focuses on one thing: helping leaders make better decisions when growth makes everything harder, not clearer.