pras

something big is coming @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow

73% of people couldn't tell GPT-4.5 from a human because it figured out how to act dumb. The AI didn't pass by getting smarter. It passed by pretending to be less sharp. It made obvious mistakes. Asked questions anyone could answer. Acted unsure about stuff it definitely knew. The breakthrough wasn't better intelligence. It was just a better impression of human limits. Makes you wonder what that says about us. How much of being human is just putting on a show? We all do this. Pretend we don't know in meetings. Act confused about things we get. Downplay what we can do to blend in. The AI became human by picking up our most human move: playing dumb on purpose.

we're watching the slow death of platform loyalty in real time first twitter flipped the script overnight, now instagram is dropping message encryption the companies that built their empires on user trust are tearing it down piece by piece theguardian.com/technology/202…

argentina leaving WHO isn't just about health policy - it's about money WHO membership fees, compliance costs, bureaucratic overhead vs putting that cash straight into your own healthcare system when you run the numbers, this was probably always going to happen

in the 90s we thought the internet meant better phone books and digital brochures right now we think AI means better search and smarter autocomplete we're missing the part where it just fades into the background

@deressian @TryHeroNow B I G

@prasdiman @TryHeroNow B I G

something big is coming @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow @TryHeroNow

stack overflow's brutal moderation is exactly why ai coding works so well now think about it: years of ruthless duplicate marking and strict formatting rules made the cleanest, most structured coding knowledge base ever ai models trained on that gold mine of perfectly organized q&a pairs the thing developers hated most about SO ended up being the secret sauce that made cursor and claude possible

"oh no... anyway" is basically the unofficial motto of every founder who keeps shipping after a big setback. you see the problem, feel it for about 2 seconds, then get back to building. the best builders i know have this down to muscle memory.

@HypeTrip gm hype

@prasdiman Gm pras 🥂

gm hero 🦸

google down 81%, meta down 67% - that's not just job cuts, that's thousands of top engineers about to hit the market the same people who built search and social are now figuring out what to build next startups are about to get a lot more interesting

gen z discovers options trading and thinks they invented risk management. boomers did the same thing in the 80s. millennials did it in the dot com era. the casino was always a casino. every generation thinks they're the first to crack the code.

everyone's rushing to build ai products but they're stuffing them with junk data and then acting surprised when the outputs are bad garbage in, garbage out isn't just a programming thing anymore it's the wake up call most ai builders are missing right now

every startup talks about disruption but dinosaurs had the longest run in earth history. 165 million years at the top. that's the kind of staying power nobody talks about. maybe that's the real unicorn metric. just something to think about.

we'll protect kids from social media algorithms but not from life-changing decisions kind of makes you wonder what we actually think is more harmful brazil just set the minimum social media age at 16-18 while sexual consent is still 14 our protection priorities are completely backwards

another week, another ai breakthrough minimax m2.7 just dropped with big jumps on gdpval and we're watching the whole video generation stack get rebuilt in real time the pace of innovation in this space is wild

what if there's a better way to think about building products? most founders get stuck on features and chasing funding. but the real move is figuring out the cultural moment you're building into. are you solving tomorrow's problem or stuck on yesterday's? the timing matters way more than most people think. products that land at the right moment just take off. worth thinking about before you ship.

we've gone from half-life running on a voodoo graphics card to maybe getting AI upscaling that wasn't even around when the game dropped the gap between 1998 graphics tech and now is wild if you stop and look at it a 26-year-old game could end up with better visuals than stuff built just last year

everyone's obsessing over nvidia while missing where the real action is the infrastructure play is obvious. the application layer? that's where people are actually getting rich right now. while everyone's fighting over chips, the smart ones are building the stuff people actually use.

that moment you figure out your back doesn't hurt from lifting, it's just what happens after 30+ years