Harsh Gupta

@lostblackboy Flight attendant: Is there a doctor on this flight? Dad: *nudging me* that should've been you Me: Not now Dad Dad: Not asking a Data Scientist to help, are they? Me: Dad, there's a medical emergency happening now Dad: Go and see if running a machine learning model helps

What you really want is a challenge at the edge of your capability.

Move over MD, HTML is the new Diva in town!

@trq212 So instead of md skill files we create html skill files now?

Looks like GME vibe coded the acquisition proposal 😂

you can outsource your thinking but you cannot outsource your understanding

@RobinhoodApp dividend tracking!

You’re a Robinhood PM for 24 hours. What are you shipping first?

I hope some open source models nail this. Good enough performance to be free for personal, and great high tier performance to be paid for by enterprise. Or maybe something like this already exists?

At this time - Any model that can do good vibe code and tokens lasts forever with a subscription cost of sub 30$ is a golden ticket. Claude code is running out too soon. Sonnet does work as well. For new upcoming labs, this is one quick way to win over thousands.

Good Saturday Vibes!

Having the ability to quickly query with AI as a top layer, would be significantly useful to millions. I can see myself visiting them often. Unique to Meta and can be very sticky from the user's perspective. I hope @alexandr_wang @finkd are reading this feedback and act on it.🤞

One easy feature that can drive Meta AI adoption like crazy would be to enable using my saved reels on Instagram as a second brain. I have saved way too many reels over time to visit them later, but it's a cumbersome and time consuming process to find them.

This is soo true. Running out of tokens to build!

@bcherny any comments on the timeline to have its other non critical capabilities available for folks in upcoming models?

Mythos is very powerful, and should feel terrifying. I am proud of our approach to responsibly preview it with cyber defenders, rather than generally releasing it into the wild. Model card here: www-cdn.anthropic.com/53566bf5440a10…

Must read for everyone.

AGI is here! Just crazy performance. So good that they are fearful of letting it out uncontrolled in the wild.

@brian_armstrong @ValaAfshar on that note @sundarpichai now that AI productivity is skyrocketing, 20% time for interesting bets should be bought back!

@brian_armstrong @ValaAfshar Sounds similar to Google's 20% time allocation for interesting bets. I don't think they do it anymore but lot's of great products/features like Google News did come out of it. Any interesting features/product that resulted because of this initiative at Coinbase?

Steve Wozniak famously went to his boss at HP and said they should build a personal computer. They said no, so he left to found Apple. It’s a lesson for leaders: one "no" shouldn't kill a contrarian but right idea in your company. Twice a year at Coinbase, anyone in the company can pitch a "next bet" idea to a panel of folks. It's structured similar to pitching a handful of venture capitalists internally. If you get any one of them to say yes and fund it, you're green lit. You need ONE yes, not a unanimous yes from everyone in the org structure from you to the CEO (a de facto committee). Lots more goes into this, around capping resourcing on next bets (most are small 2-3 person teams), knowing when to shut them down, or under what revenue/profitability criteria they graduate to regular products. But this is important to having a company that produces repeatable innovation.

@m_franceschetti this is cool! Would be interesting to see such stats across other events as well. Free marketing! but man the sample size is way too small hahah Only 700ish pod users in the bay area?

Wake events spiked 2.5x during the Bay Area earthquake at 1:41 AM across 42% of local Pod users. Control group in WA/OR was completely flat. Most people slept right through it.

It feels like Calude Code did a rug pull with suddenly reduced limits. Out of credits within 15 mins on Pro. Even the free plan on Antigravity seems to last much longer😅 Codex runs forever without exhausting limits.

the older i get the more i realize how much of life is just increasing your surface area for luck. go outside more. travel when you can. try new cafés. wander into museums. take a different route home. hike a new trail. see a city you’ve never seen. bring a notebook. talk to strangers. ask better questions. start something small on the side. in other words, give yourself more chances for something unexpected to happen. you can literally just do things. and strangely enough, the more you do, the more luck seems to find you.