DeepSeek's disruptive effect is a powerful example of why stories matter.

DeepSeek was different.
DeepSeek was made in China.
DeepSeek came out of nowhere.
DeepSeek was "cute."
DeepSeek broke the narrative.
DeepSeek was free.

The reality is OpenAI is still way ahead.

They have the resources, they have the talent, they have the chips, they have the developer toolkit, they have the revenue, they have the broad, mature, and quickly growing product. And o3-mini just released.

Simplifying Trump's opposition of climate change to "I think wind farms are ugly." is a shallow read of a complex situation.

The incentive is to strengthen the petrodollar.

This is important to understand. 🧵

I used to believe AI could lead to the singularity, a runaway explosion of intelligence. But after listening to some of the smartest people on the internet, I’ve changed my mind.

The first model T didn't have brakes. No seat-belts, no airbags, just an engine with a shell. It could never be sold today. But it changed the world.

We are building the engine of AI. It's a risk. But innovation means putting progress first.

The next breakthrough in AI is removing humans from the loop.

DeepSeek is revolutionary because it is self-reinforcing, with no need for human feedback.

The next hurdle is self-training AI. If it can create its own data, it won't need our help.

99% of people don't understand the hype behind DeepSeek.

It's more than just the #1 app.

It's removing humans from the loop. 🧵

SEO is dead. Most searches can be answered in a sentence. Google was fighting a losing war by rewarding long, stuffed articles. The incentive created terrible content that people didn't want. I'm excited to see if novel, compressed ideas begin to win.

Paper is the ultimate interface. We've been chasing that since the beginning. Your screen is your "desktop".

Generative UI won't have to be explained. It will be as intuitive as a crayon is to a child. You'll conjure UI by drawing a circle.

Americans have the courage to try. It makes us great. It lets us fail. We swing for the fences, risk it all, and go for broke. And sometimes, it pays off.

I can feel how valuable time is to my grandfather. The proud, loud, audaciously intelligent man he knows himself to be is being lost to time. Someday we may not recognize each other. It is a silent pain we both bear, but it crushes him while I still walk.

Listening is the most underrated skill.

Bad code reduces complexity.

Good code expresses complexity.

Love is the only way to reclaim your power.

AI agents aren’t just a buzzword—they’re the solution to AI’s toughest challenges and the key to its future.

I’m taking the leap. After years of preparation, I’m betting everything on my dream. I’m going full-time into my startup, Celest with 6 months of runway.

I'll be building in public, sharing the wins, the struggles, and the lessons, so others can see what founding a startup looks like.