SimplyHuman

San Francisco, California

@adxtyahq Fuck you dude. .. clickbait assole

Wait… Gemini 3.1 Pro just one-shotted this insane CSS animation???

@AshikArtistry Bait bait bait

Gemini 3.1 pro btw One shot, design We're cooked for real!

@MengTo You don't have the prompts. Just a clickbaiy

Gemini 3.1 Pro is so freaking good at making skeuomorphic user interfaces and animating them

@rohindhar is it a good time to buy in SF? miraloma park?

446 homes for sale in San Francisco right now, compared to 674 this time last year (Down 34%)

@danmartell Your tweets are the best 🤩

If you really cared about being the best in the world, you'd go to sleep. You gotta sleep on things to get better at them.

Watching F1 on a flight and it hit me: startups are basically race weekends. Practice = prototypes. Qualifying = distribution. Race = retention. The winners aren’t the ones with the fanciest engine… they’re the ones who iterate fastest between laps.

Personal thing I’ve learned working in AI research while trying to start a startup: I’m way better at building models than picking what to build 😅 So my rule now is simple: ship something tiny every week and let users tell me what’s worth scaling.

AI startups don’t win because they picked the “best model.” They win because they picked a painful workflow + defined “correct” + built evals like CI. If you can’t measure quality, you don’t have an AI product — you have a demo.

Jensen Huang this morning: “Claude is incredible. Anthropic made a huge leap in coding and reasoning. Nvidia uses it all over. Every software company needs to use it.” claude code cooked

Seriously, where the heck is Meta and what did they do with all those expensive Aquihires? So far, absolute silence, not a word about upcoming releases or products. This is insane.

here's a random thought: everyone wants to build the next ChatGPT but most AI researchers at big tech companies can't even ship a simple chrome extension the problem isn't technical knowledge it's actually just being willing to build something rough and let real people try it - start with one tiny AI feature - ship it this weekend - watch people actually use it - change things based on what you hear that's how you go from research to startup stop perfecting in the lab and start failing in public

Building AI research models is weird. You spend 90% of your time on stuff users never see: - tuning hyperparameters - cleaning training data - debugging obscure edge cases But that invisible 90% is what makes the visible 10% feel "smart" and seamless to users 🫡

@damienghader PROMPTS

@damienghader PROMPT

AI researchers & Google SWE's What AI projects are you exploring now? Share your cutting-edge models below, join 100+ innovators here every week. #MachineLearning #buildinpublic

Starting 2026 with high hopes then AI throws its usual curveballs feels like riding the wave while trying not to drown in the tech chaos