If All You Know Is the Carriage, You'll Never Reach Hawaii

Nagamura: EVeM is a company that provides training programs for managers at startups. I used to think we were in a line of work that wasn't much affected by labor or technology, but we're now at a phase where we're trying to build AI into our product and grow the business with it.

Specifically, we were trying to take the management support our human coaches provide and turn it into AI—and in the middle of that, AI was evolving at a dizzying pace.

I often use this "carriage and Hawaii" analogy. AI is basically a means to an end. But sometimes the means swallows the end. If all you have is a horse-drawn carriage, the goal of "let's take a trip to Hawaii" never even occurs to you, right? It's because the airplane was invented that the goal "let's go to Hawaii" can exist at all.

In other words, if you don't know how the means is evolving, you can't even come up with good goals. That fear grew stronger by the day. "Am I really painting an accurate picture of what's possible for this business?" A leader who doesn't understand AI might look to employees exactly like a boss who keeps insisting, "There's no way we can ever go overseas."

Nagamura: We had. I tried to get our in-house engineers to teach me tools like Cursor, but it didn't work out.

For one thing, our people are busy, and this wasn't their dedicated job. Making materials for me, figuring out the right order to teach things in, putting together a program—all of that is a huge burden in itself. In the end it became a "teach me on the side of your day job" arrangement, and the project just fizzled out partway through.

So then, should I just do it myself? Honestly, I felt out of my depth (laughs). I'm more of an analog person—I'm far more interested in things like finding the right words or observing how people behave. I just couldn't get myself into the world of technology.

Feeling out of my depth, I couldn't get good at it; not getting good at it made me feel even more out of my depth. A complete vicious cycle. I knew I had to do it, but I just couldn't bring myself to start—and that went on for a long time.

And this is just my own hunch, but I think everyone's actually just bluffing about AI (laughs). Look at X and it seems like the whole world has mastered AI, but I doubt more than a tiny fraction are really using it properly. People are just too embarrassed to admit they don't know. I really don't think everyone has genuinely caught up.

Nagamura: First off, as a business-side person, I had this "engineer complex." When I was at an IT company, I was in the position of selling the finished product without understanding the process by which it was built. When I talked with engineers about that process, it was so specialized that I couldn't keep up. I also have this complex where I get embarrassed when I'm confronted with something I don't know—so I felt a little guilty about working with engineers.

On that front, it mattered a lot that Wolkin was an engineer who was so easy to talk to. And something I think is really important: not once was I told, "That's no good." Not a single "you didn't manage it" or "that's not enough."

This is something I value when I work with people as an advisor too: when you're with someone who only knows 1 out of 100, you don't say "no good" from the vantage point of knowing all the other 99. Wolkin never did that. No matter how small the step, they responded with encouraging communication. For me, that was a really good thing.

Nagamura: The flow was a lecture and dialogue first, then homework.

For me, getting to start with a lecture was great. If the style had been to jump straight into hands-on work, it would have been pretty rough on me emotionally. Having time where I could just listen at first was honestly a relief.

It also helped enormously that I could rewatch the recordings. When I sat down to do the homework, I'd already forgotten what the lecture covered. So before tackling each assignment, I'd rewatch all the recordings first, every time.

And above all, the way the homework was set was exquisite. Technically you could complete it just by following the instructions, but the design subtly required you to figure some things out for yourself.

For example, the first assignment was just to put a prompt into ChatGPT and drop the output into Cursor. That's the first step. But by the second assignment, there's hardly any step-by-step guidance—it's more like "try building this kind of app." The first, second, and third assignments ramp up in difficulty just right. I really thought the way the homework was designed was masterful.

Nagamura: Once I tried just whipping up a web app and putting it out there, other employees started saying things like "I'll try making a landing page myself." We're a small company, so it pretty much comes down to what the CEO is doing. I think the CEO actually doing it himself was the single biggest trigger for changing the mindset inside the company.

I also built a diagnostic app myself, and it was fascinating. Things I couldn't see when I was doing something similar in a spreadsheet became visible once I turned it into a working app. The quality of the inputs that came in was completely different too.

I think the very fact that "I can build something that works" expanded the organization's creative thinking. As a company, too, I really feel our standard has risen a lot—from just chatting with ChatGPT to building a proper app yourself and shipping it quickly.

Nagamura: The biggest lesson was the change in mindset. Before, I thought "I have to study like crazy to catch up with AI," but I came to a stance of "I can just ask AI about that part too."

It wasn't about studying like crazy—they taught me the feeling of "playing with AI." Even when I'm building something in Cursor, I can ask ChatGPT first, drop the output into Cursor, and sometimes it builds in one shot. It doesn't always go smoothly, but even that gave me the feeling of playing a game, which I hadn't felt in ages. It was so much fun I ended up building my own chatbot, and honestly, I could barely sleep because I was having so much fun (laughs).

That said, AI evolves so fast, doesn't it? I'd finally gotten the hang of Cursor, and then along comes Claude Code. AI agents talking to each other, and all these other baffling things keep coming up. Normally you'd think, "Ugh, I have to learn something new again," and honestly it gets to be a hassle.

But in moments like that, Wolkin is right there beside me. They invite me with "Hey, let's play around with this a bit," and give me a little assignment. And once I try it, it gets really fun again. No matter how fast the world of AI changes, having someone who says "let's play with it" each time is incredibly reassuring.