The On-Ramp Problem Nobody Is Building For

The On-Ramp Problem

8:46

The On-Ramp Problem Nobody Is Building For

I'm dictating this through an earpiece. Not because it's novel. Not to make a point. It's just become how I work. At some point in the last year, my primary interface shifted away from a monitor, a mouse, and a keyboard toward something closer to a conversation that follows me from device to device. I get responses back in audio. I think out loud. The model listens, processes, and keeps up. It fits inside the life I already have and that shift — mundane as it sounds — is the most important design insight in AI right now. Not the model capabilities. Not the benchmarks or the reasoning scores. The interface.

We've Been Building for Ourselves

I've written before about the 84-percent problem — after three years of breathless coverage, roughly 6.8 billion people have never typed a single AI prompt. Not once. And the companies who figure out how to reach those people will define the next decade.

But I left something out of that piece: why the grey dots are still grey.

It's not access. Smartphones are everywhere. ChatGPT is free. The barriers to entry are lower than they've ever been. It's the on-ramp.

The builders — the red dots, the 0.04 percent — have spent four years building AI for themselves. Tools that reward curiosity. Systems that respond to experimentation. Interfaces that assume you already know what a context window is, what an agent loop does, what it means to iterate. The fluent operators who can now move billions as a team of one got there by drinking from a fire hose for years. By playing. By failing, adjusting, and building intuition that has no manual.

That's not a scalable on-ramp. That's a self-selected filter.

The result is exactly what you'd expect: everyone around you is using AI at the lowest possible level — because how would they know they can do more? Nobody told them. The tool didn't meet them where they are. And the interface didn't show them what's on the other side of the wall.

The Google Moment We're Still Waiting For

Here's the thing about Google. Most people don't know what a server is. They don't know how search indexing works. They don't have opinions about SEO. They just know: type what you want, and you'll find it.

That's the bar. Not literacy. Not training. Not a certification. Just a thing that works the way people already think. AI hasn't found that moment yet. The cloud models — ChatGPT, Claude, Gemini — got closest. You talk, it talks back. That's intuitive. That's why those are the ones that got adopted. Not because they're the most powerful. Because they removed the most friction. But even that interface is still built for people comfortable with a blank text box and the vague assignment to "say something." And most people are not.

The Google moment for AI looks more like what I've stumbled into — voice-first, ambient, embedded in the devices you already use. You don't open an app. You don't stare at a cursor. You just talk, the way you already talk, and something smart talks back. The model comes to you.

iOS 27 might be part of the answer. Baking AI into the OS layer — not as an app you launch but as a thread running through everything — is the right architectural instinct. When the assistant knows your calendar, your habits, your voice, and it lives in your ear rather than on your screen, the interface becomes invisible. Invisible is the goal. The best technology disappears. The map became GPS became a voice that says "turn left in 200 feet." Each step made the tool smaller, quieter, more embedded in the moment you're already in. AI needs to finish that arc.

The Stack Problem

NVIDIA made noise recently about putting a GPU in every home. I love the ambition. The concept of a personal AI stack — your model, your data, your compute, sovereign and local and yours — is philosophically exactly right.

But here's the honest problem: that works for me. It does not work for most people. A personal AI stack, as currently conceived, requires comfort with configuration. With tradeoffs. With the idea that "managing your model" is something a normal person would want to do. And the vast majority of people — brilliant, capable people who have simply chosen different paths — are not going to get there through documentation.

What they need is something that gives them the feeling of ownership without the burden of administration. Something that makes the data feel like theirs, the model feel like it knows them, and the interface feel like a conversation rather than a command line. Something where the on-ramp is so natural they don't even notice they're on it.

We don't have that yet. What we have is a gap — between the cloud models that are easy but feel extractive, and the personal stacks that are sovereign but feel technical. That gap is the product problem nobody has fully solved.

The Curiosity Problem

There's one more thing that keeps coming back, and it's harder to solve than the interface. We've edited curiosity out of our culture.

The people getting the most out of AI right now got there by playing. By asking weird questions. By seeing what happened when they pushed in unexpected directions. That open-ended tinkering — trying something not because you know it'll work, but because you want to find out — is exactly the disposition that unlocks AI's real capability. It's also the disposition that formal education spent years training out of most people. The right answer matters. The rubric matters. Coloring outside the lines is a problem to correct. AI rewards the opposite instinct. It rewards the person who says I wonder what happens if I try this more than the person who memorized the approved framework.

I wrote about this in the identity inertia piece — most of us built our professional identity around a role, a title, a fixed definition. And AI is dissolving those faster than people can rewrite their résumés. The people with the most to gain right now aren't necessarily the youngest — it's everyone sitting on twenty years of pattern recognition and experience, which is exactly what AI amplifies best. The tragedy would be watching decades of insight go underused because of a little technological discomfort.

This Is a Permission Slip

We're living in a strange, beautiful time. A time where a thought can become a thing overnight. Not through teams, or funding, or decades of engineering — but through collaboration with machines that think just enough to amplify us. It feels unreal. But it's not. It might be the most human thing we've ever done. Because at our core, that's who we are. We are creators. If you cracked open the source code of humanity, the header on that file would read: Made in the image of the Creator. It's written in us — in our instinct to build, to imagine, to fix, to form, to make. And for the first time in history, the tools have finally caught up to the blueprint.

AI isn't replacing creativity. It's revealing it. It's the bridge between imagination and execution — the amplifier of human will. The magic isn't that the tools can do so much. It's that you can now do more than you ever thought possible. I'm not writing this for the engineers or the builders already in deep. I'm writing this for the ones looking at all of it and saying: that's amazing, but I could never do that. You're wrong.

This isn't about talent or training. It's about truth. You were made to create. The spark you feel when an idea clicks — that's not accident. That's design calling you home. This isn't a moment for a few gifted people. This is a global permission slip. A collective awakening. A rediscovery of something that's been inside you the whole time. The on-ramp is still being built. But you don't need to wait for the perfect interface to walk through the door. The tools are waiting for you.

Rich Washburn is a technologist and strategist based in Fort Lauderdale. He works at the intersection of AI infrastructure, national security, and capital. Managing Partner and Chief AI Officer, Eliakim Capital.