Writing · May 2026

Two Months In. Here's What I Actually Learned.

The last post was about how to get started — what to build first, how to make your first project happen.

Two months later, what's worth writing about is something different. Not tactics. The stuff underneath.

Protect your inspiration

Inspiration isn't something you wait for, but you can protect the conditions that let it show up.

The best ideas never come when you're sitting at your computer trying to have them.

People ask where my product ideas come from. The honest answer: always somewhere unexpected.

Understory — a web art project exploring words in other languages that English doesn't have — started at a hair salon. I was sitting there with nothing to do, completely zoned out, and scrolled past a post about the Japanese word komorebi: the way sunlight filters through leaves. Something clicked. Could you make something that lets people actually feel these words, instead of just reading them?

SmokeStory — a tool that turns NASA satellite data into wildfire stories humans can actually read — started with a conversation. I was catching up with a Yale classmate who does remote sensing. I'd loved that kind of data back at WRI but hadn't touched it in years. After we talked, I started thinking: in all the satellite data I've seen recently, what's moved me the most? The first image that came to mind was the California fires. From that image to a product, there was really just one step.

Once you're constantly in problem-solving mode, that struck-by-lightning feeling stops coming.

Inspiration comes from your whole life — whether you have enough empty space in your days, whether you're touching things that have nothing to do with AI, whether your nervous system has actually gotten to rest.

Your “why” is your actual moat

Why you build is harder to copy than what you build or how you build it.

Every product I've made started without a plan. Looking back, they've naturally fallen into three categories. The first is Climate Intelligence: SmokeStory, True Zero. These draw directly from years of judgment I built at WRI and BCG — that foundation is genuinely hard to replicate. The second is Workplace & Personal Tools: NoThanks, my personal site. The third is Web Art & Learning: Echo, Understory, and an ongoing public record of how I'm figuring this out.

None of it was planned. Each one started with a sudden impulse to make something.

I've come to accept that using AI for pure self-expression is completely valid. After reading Project Hail Mary, I wanted to hold onto that feeling — cold, vast, quiet but not lonely. Before AI, I might have posted something on Douban or in my WeChat Moments. Instead, I built Echo — a starfield that breathes with music, so other people could feel that same feeling. That's a completely new form of self-expression that AI made possible for me.

The things you build spontaneously are the most authentically you.

What drives you to make something — your instincts, your curiosity, the judgment you've built up in a particular area over years — that's the part that can't be copied. Passion isn't a strategy, but it is a moat.

Nobody has the answers. That's fine.

Finding out that everyone is figuring it out too was actually a relief.

Everyone is feeling their way through this — including the people who look like they have it figured out. I went to a Yale alumni event once, genuinely hoping to find answers. Nobody had them. Not because they weren't sharp, but because there simply aren't definitive answers to these questions right now. Everyone has their own version of the anxiety; it just looks different on the surface.

For a while I felt like I was being pulled in six directions at once. Hardware or software? Start monetizing? How? Every time I took in too much outside input, my own voice went quiet — and so did my ideas.

Sometimes the only way to start outputting is to stop taking in so much.

Know yourself. Then move first.

The differentiator isn't just who you are — it's whether you act from that place, or wait for someone else to tell you it's worth doing.

When everyone can vibe code, the reason someone picks you isn't your tools or your output speed. It's you — your taste, your judgment, the particular way you see things. That's not something AI replicates.

Knowing what you're good at doesn't do anything on its own. Agency is what turns it into something real.

My “AI Activation” service wasn't the result of market research or competitive analysis. A friend asked if I could help them get started with AI. I walked them through it. Then more friends asked. I eventually realized: this comes naturally to me because it's already my strength — not something I built deliberately.

Most of what I've built had no clear market signal when I started. I just wanted to make it. Not because it was trending. Not because someone validated the idea first. That impulse, followed — that's what makes it yours.

The answer was never something you find outside. It grows from within — and you have to be the one who acts on it.

Two months in. There's still something new I don't know every single day. But the unknowns are getting more specific — not “I have no idea what AI can do,” but “this particular question, I haven't thought through yet.” That specificity is its own kind of progress.