How to be a builder

As a kid of working-class immigrant parents in New York, money was a tough topic. Did I ever doubt I would lose my home? No, fortunately not. Did I ever worry about being too underdressed to brave the wintry streets for the walk to school? Absolutely not. Did I always eat enough food and easily access cut up fruit? Every day. My parents made enough money to keep the family healthy and living. But there was not much money leftover for dreams. And by 1998, I was dreaming about the Internet, a lot.

Thanks to the massive direct-mail marketing campaign from AOL, it was inescapable. I had some exposure to computers from school and I was absolutely hooked. Our mailbox was filling up with catalogs from HP, Compaq, Dell. And I really wanted a home computer to message my friends on AIM. Some of those catalogs advertised computer parts, and flipping through the pages of TigerDirect and comparing the different specifications of motherboards and RAM and hard drives, I realized something. These parts put together were significantly lower cost than the sticker price of a Compaq or HP. If I wanted a computer that was comparable to one of these systems—no, a computer that was better spec'ed than those systems–I needed to work on something different. I needed to build a computer.

That is the work that kicked off my life's work.

There is a lot of discourse with AI tools that they make everyone a builder. And there is even more ominous message that anyone who does not adopt AI tools will fall behind and never recover. The outcome will probably fall somewhere within the zone of overhyped and underestimated. Skepticism of AI that warrants late adoption is perfectly reasonable. But underestimating AI would be a mistake, not only because how much is being built to support its development, or how much is being built with it, but simply how much it can teach on what it means to build.

Learning how to build computers grounded my understanding of hardware, software, and networks. Before there was cloud infrastructure, there were server farms. Before there was a server room, there was a NAS. When I first started using generated "reasoning" in 2023, I thought ChatGPT was clever for menial tasks like meal planning and fitness routines. When LLMs evolved into contained programs for executing work and the word "agents" took on an entirely new meaning—that was when my intrigue shifted. Suddenly, there was an entirely new system to build. Singularity has entered the chat. 🤪

I am positively humming with thrill to build again.

I won't pretend that building is not intimidating. When I had the parts assembled for my first-ever personal computer, I diligently wore an anti-static wrist strap and touched the computer case before lifting any component out of the bag. It was absolutely terrifying to consider that I could fry a couple hundred dollars of equipment because of my desire to experiment and learn. You're allowed to think. I kept going because finishing the build was important. Dreams take work.

It is important to understand why you want to build to persist through the moments of uncertainty, because the ground truths are shifting fast. Many industries are changing because of the technology shift brought by GPTs, but the change is still very early stage. The early-stage gamble needs to feel energizing because it is going to require a lot of energy to separate the value and the slop. I would not recommend learning and building on AI if the primary drivers are fear and pressure because those lessons are not enduring.

Figuring out what are you dreaming takes work too, and more likely to survive the outcome of a dream. Want to be a builder? Build the dream.

The Knicks spent roughly 53 years building and working on that dream. There were many moments the dream was close to feeling shattered. And then, it might feel like overcoming 99.6% is suffocating and unfathomable.

But the work that lives in the 0.4% becomes something you carry for a lifetime.