Lessons on Startups
I'm learning hard lessons while building Celest, an AI-driven software for content creation. My goal is to create an interface that AI can use alongside a user. It's pushing the limits of my ability and helping me grow—both as an engineer and as a person. Here's what I've learned:
§Projects are always 80% complete
The finish line always feels close, but it never quite arrives. You are building something that will evolve and grow. No matter how many deadlines you set, hit, or miss—the work remains constant.
§The last 10% takes 90% of your effort
The closer you get to finishing, the harder it becomes. The beginning of a project feels great because there's no resistance. As it grows, everything must be built in consideration of everything else. The friction compounds. The edge of a problem is sparse and wide—and it takes effort to traverse.
§Composition is powerful
I've stopped chasing perfect abstractions. Now I create flexible, reusable blocks of code and compose those blocks into unique patterns. I relate this to making music: twelve simple notes can combine into nearly infinite arrangements of chords and melodies. Their beauty lies in simplicity and harmony. Great code is the same—it's simple enough to enable powerful combinations, and expressive enough to handle complexity.
§Don't reinvent the wheel
Many common challenges have already been solved. If you need something to roll, and a wheel exists, use it. Don't waste your talent—go build something new.
§Always improve the factory
Construct a factory that produces high-quality products: modular, extensible, and easy to upgrade. Actively keep the factory clean, organized, and up to date. This is the kind of work that compounds. It can be reused across projects, refined over time, and give you a real edge over your competitors.
§Be a perfectionist
Focus on your core product offering, and perfect it. Your first impression defines your brand. It has to be unforgettable. Stay focused on the single aha moment you need to create.
§Doing is knowing
Conceptual understanding is no substitute for practical experience. Knowledge lives in the mind until it is practiced—then it becomes embedded as skill. Doing humbles your tongue and sharpens your perception.
§Do until it's done
There's a lot that goes into building a production-grade application. There is no shortcut to doing it well—it may take years. Make sure your lifestyle and circumstances support that kind of long-term effort.
§Do it your way
Surrender to your own way of being and find your optimal flow. Everyone works differently, and no book can teach you how to be you. The greatest things are created by people relentlessly pursuing their interests. Your path is guaranteed to be unique. The important part is getting started. Start building. Start learning. Start improving every area of your life.
The goal is to live a good life. Building a great business is just a part of that.