I spent 80 minutes with Henry and Chris on The Greek Plug Podcast. We talked about growing up curious in Athens, the startup I had to shut down in 2012, building a company across time zones, Nietzsche, and why I asked an AI model how to build an Iron Man suit. The thread running through all of it was the same: in the age of AI, the skill that matters most is learning how to learn.

Here are the five ideas I’d keep, with timestamps if you want the unpolished version.
Get new essays on APIs, the agentic web and engineering leadership by email.
1. Getting an answer is not learning
You can now ask any question and get a fluent answer in seconds. That is useful, and it is not the same thing as knowing something.
“Asking and getting answers is not learning. Just because you got an answer about ancient history doesn’t mean you know ancient history.”
Real learning still feels like marathon training: it hurts at the start, and the pain is the part that changes you. (08:30)
2. Go back to the beginner’s mind on purpose
I keep putting myself back at the beginner’s mind, because that is when new things start happening for me. For a CTO that means being willing to feel like the intern again. It is uncomfortable, and it is the only reliable way I’ve found to stay relevant when the tools change every few months. (09:50)
3. Traction you can’t pay for is just a bigger bill
In 2012, as university students, we built a social network that was getting real traffic in Greece. The local funds of the time wanted e-shops, not us, and we couldn’t pay for the servers. “I remember the day we shut down the servers.” The lesson I took isn’t “fail fast.” It is that the business model has to pay for the thing you built. (15:20)
4. Skip straight to a startup and learn to be an operator
When students ask me whether to do a master’s or join a startup, my answer is the startup. “You don’t learn just to be a developer. You learn how to be an operator.” You do marketing, sales and ops, and you take the trash out yourself. (16:55)
5. Do stupid things with AI, and study science
“I’m doing a lot of stupid things with AI to understand the limits, literally every day.” I keep a list of problems it couldn’t solve and try them again. I even asked a frontier model what it would take to build an Iron Man suit. The bottleneck it found was energy, which is exactly the kind of problem you don’t solve by optimizing. You solve it with science. So when people ask what our kids should study, my one-word answer is science, plus some philosophy, because we run on the same hardware as people did three thousand years ago. (35:00, 53:05, 1:05:45)
Keep going
Thanks to Henry and Chris for a conversation that went everywhere, including souvlaki versus kalamaki and where Atlantis really was. More talks and interviews are on my Speaking & Media page. If you’re new here, start here, or read how I measure AI ROI in hours returned.
What’s the last thing you learned the hard way, when you could have just asked for the answer?




Leave a Reply