Watched the digital yellow brick road at the Sphere as Google Cloud closed out their exclusive event. Live orchestration and real-time generation without a single pre-rendered scene. Crossed paths with Francis deSouza (COO of Google Cloud) as well. A fitting finale to a week

I had the opportunity to sit down for lunch with Malcolm Gladwell during Google Cloud Next ahead of his keynote. Our conversation focused on how ideas spread, why adoption stalls, and what it takes to make change stick inside complex systems. A few other points stood out to me from

This is an excellent example of why it's vital to ensure everyone has tools to audit and customize intelligence, not just a handful of frontier labs.

AI policy cannot be written solely by the people set to profit from deployment. We already saw what happens when economic incentives dominate policy; the public interest disappears from the conversation. Models that decide credit limits or medical triage require input from

while all our friends were at presidio for memorial day, my cto was fixing a deployment for a fortune 20 client this is why you will lose, and we will win.

>wrote down "I will be the CEO of Intel" in his 20s you can just do things.

over 500 people want in to our YC Startup School afters, >20% of all attendees we rejected 85%+ of them keeping the bar extremely high is essential going to be a night to remember 👀

watch out TBPN 👀

never seen a group of more cracked students than at our AI startup school afters 2 IOI gold medalists, 3 IMO gold/silver medalists, Putnam fellows…unreal thanks Y Combinator and Gradient for helping put together an amazing night

Only CTGT knows what's "actually working in enterprise AI" Everything else is low RoI, smoke and mirrors, or a combination of both venturebeat.com/ai/announcing-…

News: CTGT wins Best Presentation Style award at VB Transform 2025 venturebeat.com/ai/ctgt-wins-b… via VentureBeat

no bbq and parties, this is how july 4th looked for our cto at 5am this morning when a fortune 20 co drops a feature request, you have to skip out on everything your friends are doing people talk a lot about the highs of being a founder, but skip over the quieter moments

Regulation of AI is about shaping policy around a government's values. Look at the U.S. vs. China: 🇨🇳 China’s approach: AI as a tool to reinforce state narratives. 🇺🇸 U.S. approach: AI as an engine for questioning, discovery, and innovation. People think the AI arms race is

I shared a Stripe Sessions dinner table with leaders whose firms collectively top $1T in market cap, co-hosted with Denise Teng (Gradient at Google). Mission: figure out how to deploy AI that actually works in production. 5 essentials we landed on for moving from demo

Startups win on narrative, not code. That’s what stood out to me at the Character Capital Workshop with John Zeratsky and Jake Knapp. We talked about how a conference paper on esoteric math turned into a seed round led by Google and engagements with some of the largest companies on earth, by

AI will trigger an intellectual renaissance. Or it will trap us in dopamine-driven content loops. The difference isn't in the tech itself, but how we choose to use it. We need AI that fuels critical thinking, discovery, and creativity, not just addictive algorithms. The big

Wall Street has stopped calling AI a toy. It’s now parked on the P&L right beside interest expense. I had a panel alongside sessions with Jamie Dimon (CEO, J.P. Morgan), Treasury Secretary Scott Bessent (U.S. Secretary of the Treasury), and a roomful of investors at J.P. Morgan’s seminar during

Demis always knew the path to AGI is through games

“Bigger model” bragging rights are done. The next fight is proving that AI can be trusted and pay its own bills. At the Deepwater AI Summit panel with AJ Christensen (Head of Product, Fastino) and moderated by Mike Olson and Gene Munster (both Deepwater Asset Management), we asked: one

So the VP of an Alabama credit union, the CEO of a public Aussie fintech, a Navy AI scientist, and a Wall Street deal-maker walk into a bar to talk enterprise AI. First takeaway: a model that can’t show which data points shaped its answer never makes it past audit. Glass, not