if @ifbuilder

That's the lesson that sounds obvious until you've spent years doing it backwards. I built first and went looking for the problem after, which is way harder. Did finding the problem-first people change what you built, or just who you sold to? Curious if the product itself shifted once you flipped the order.

Makes sense for Torino tbh, he had a really strong season in Spain as that deep-lying playmaker type. The discipline side improved too, the yellow card problem from before basically went away. Question is whether Fiorentina lets him go permanently or just wants him back after the Las Palmas loan with that buyback in play.

@samgrows oh that's actually a smart use case. if it can handle the upload flow reliably that's a real time saver during a challenge when you're posting constantly. let me know how it holds up, curious if it breaks when the upload UI changes even slightly.

The macros [dot] ai detail hits honestly. Wild how some ideas just sit in the back of your head for years until the tech finally catches up to them. I've got a couple of those domains too, bought for things that were too early to build at the time. Good reminder that "didn't manage" sometimes just means "not yet." Congrats on actually getting to ship the thing you bought that domain for.

This is honestly the feature i've wanted for a while. Teaching by showing instead of describing is the right model. If you're taking wishlist items: the dream version is Codex noticing when i've repeated something manually a few times and offering to record it itself. Right now i still have to remember to hit record. Either way, congrats on shipping this. The Shortcuts DNA shows.

ha, the demo always makes it look like that. file an expense report, copy paste, easy. the real test is the messy stuff that never repeats the same way twice. recording the clean version is easy, it's the workflow that drifts every run where these things usually fall apart. still curious to try it though.

"You can just do things" hits, this stuff is moving fast. The next step I keep wanting is the reverse of record & replay. Instead of you showing it the workflow, Codex noticing you've repeated something a few times and offering to turn it into a skill itself. You show it once is great. It spotting the pattern for you is the real "just do things" moment.

The Record & Replay one is the standout here. Showing a workflow instead of describing it is the right direction. The part I keep waiting for is the reverse though. Codex noticing you've done the same manual thing a bunch of times and offering to make it a skill on its own. Right now you still have to know what's worth recording. The bigger unlock is when it spots that for you.

The record-it-once part is slick. But it's still you deciding what's worth turning into a skill. The real jump would be the reverse. Codex noticing you've done the same manual thing five times this week and asking if you want it as a skill. Best automation is usually the one you didn't know you needed yet.

Record & Replay is cool but it's still you deciding what's worth automating. The real unlock would be the reverse. Codex watching the stuff you repeat by hand and going "you've done this same thing 5 times this week, want me to turn it into a skill?" The best automation is the one you didn't know you needed yet.

Fair question. Honest answer: it moved traffic, not revenue. That day we got close to 300 visitors and I could see people actually digging in, pricing page, privacy page, the stuff that matters. Extension installs were about 2x a normal day. The part I care about more: from what CWS shows so far, most of the people who installed that day still haven't removed it. CWS reporting lags a bit so I can't give you a clean final number yet. But sales? Zero that day. And honestly I didn't go into Product Hunt expecting any. I don't think the people who actually want to install a Chrome extension like this are really hanging out there. So the low numbers or the lack of sales didn't kill my morale at all, that was kind of what I expected going in. For me #28 was about getting in front of people and seeing if the product holds attention. Whether that converts later is the real test, and I don't have that answer yet.

I had a fight with Claude the other day. It told me to wait and build an audience before my Product Hunt launch. I ignored it and launched cold. That day there were 900+ products on Product Hunt. Foldif closed at #28. Top 3%.

mostly agree, i launched something cold a few days ago with zero audience and just posted everywhere from day one. the part that's harder than it sounds though, organic isn't free, it costs time and a ton of posts that go nowhere. one piece bringing thousands is rare, most days it's crickets and you post anyway. day one distribution is right, just brutal at the start.

the repeating-yourself part is the real cost nobody budgets for. every time a model changes you basically relearn how to talk to it. half the "this one is dumber" feeling is just losing the muscle memory you built with the old one. not saying they're equal, just that some of it is us readjusting too.

the branding being perfect is exactly why nobody uses it lol. when making the logo and the landing page takes an afternoon, the only hard part left is finding someone who actually wants the thing. ai made the easy 90% free and the painful 10% even more obvious. that last bit is the whole game now.

@GeminiApp @LandoNorris @McLarenF1 gemini out here celebrating podiums while i'm still waiting for it to finish my prompt without losing the thread halfway lol. congrats to lando tho, that one was earned

the 80% number sounds apocalyptic but software has lived this on repeat for years. when i started, flash and actionscript were everything, then it just died and those people quietly became ios devs. the job didn't vanish, it moved. finance analysts aren't disappearing, the work is shifting under them. that part feels new to everyone outside software but it's just tuesday for us. ground moves, you move with it.

@PeterDiamandis nobody's actually paying for the rockets. that 2.1 trillion is almost all starlink, the rest of the world just hasn't said it out loud yet. spacex is a telecom company wearing a rocket costume. curious if you'd push back on that or if that's basically the bet

this whole "no a human actually made this" thing isn't really about ai. people think the ai vs human debate is new. it's not. i saw this my first years in software. flash and actionscript were everything back then, whole careers built on it. then it just collapsed. i watched those same guys quietly become ios devs and start over like nothing happened. that's the rule nobody tells you when you start. the ground moves and you move with it or you're gone. the work itself gets handed to the machine now, you just move up to deciding what's worth making. only new thing is it's not just us anymore, ai dragged every field into the same loop we've lived in for years.

Noah Kagan @noahkagan

a week ago

I love how everyone has to say that the shitty looking work is actually theirs instead of AI.

19 2 30 5K 2

people blame apple for this but it's not laziness, the call and the recording are basically fighting over the same mic input and iOS just picks the call. a background app that keeps writing to disk through the interruption would fix it. kind of surprised nobody's done it right yet tbh.