How we can use AI in government:

1/ When YouTube started in 2005, broadband was just coming online. It wasn’t clear the internet could even support playing video in a browser. The scrappy team hacked together novel innovations that scaled to occupy 30% of the internet’s traffic in a matter of months.

It's remarkable how many successful companies and professionals know how to present themselves in the world - through office design, clothes, templates, etc. - but still look like crap on video. Being good at video is this decade's superpower, and we want everyone to have it!

Yeah. It's nice to have a name you can spell.

Thank you Mario Gabriele 🦊 for the opportunity to reflect on two decades in VC and share some of Sequoia Capital’s philosophy on building enduring companies. For those working on the next generation of transformational companies, I hope our conversation provides some useful perspectives.

Can anyone get ChatGPT, Gemini, or Grok to make a photo of two people shaking hands with their left hand? I can’t after hours of trying.

Sometimes I wonder if the AI text spam bots even know me at all.

Interesting spam tactic: text me like *I’m* the LLM.

AI is forcing us through a series of algebraic simplifications: In many human-to-human interactions, AI being used to avoid work on one side can be cancelled by AI being used to avoid work on the other. The equation left over is what was really important in the first place.

This is good stuff.

This feels like a good time to reintroduce Libin’s Corollary, where Moore meets Murphy: “The number of things that can go wrong will double every year.”

Elon ended the Grok 4 presentation last night with, “So long and thanks for all the fish.” That’s what the dolphins tell humanity when they leave earth right before the planet explodes in Hitchhiker’s Guide. Nerds my age sometimes say this as a bittersweet parting line. I’ve

My p(doom) remains close to zero. My p(dystopia) and p(utopia) both climbing quickly.

[EVERNOTE] We released 100 improvements in 2024. We released 110 improvements in 2025 (so far!). It was tough to choose, but here are our top highlights. 👇

Alignment was the fuzziest concept in D&D and now it’s the fuzziest concept in LLMs. Let’s just call them chaotic neutral and move on to something more rigorous.

We should all have a moment of loudness. bbc.com/news/live/cn0q…

If you think AI is not gonna take your job, you might be right. If you think AI is gonna take your job, you’re definitely right.