make ideas into beautiful, consistent designs
Introducing https://t.co/vDmwlC78nr
describe your app or website once, get complete multi-screen design flows that share the same look, plus polished previews ready to share and export
the future of design is here 🔥
@kamran_khan027 Going from a terminal script to a live app on your own domain in five days is a solid clip. Curious how the summarizing holds up on long pages, are you chunking them or feeding the whole thing? And does it dedupe when you save the same link twice?
@EbubeOkonkwo26 Nice that the whole chain actually runs end to end, that's a lot of steps to keep glued together. How's the priority classification working, keyword rules or is the model scoring it? That's the bit I'd expect to drift first.
@bluemagikSGO A user asking for ads is wild, usually that means they'd rather watch a video than pay. I'd read it as a pricing signal more than an ads request. Is MyBabe paid-only right now, or is there a free tier?
@FelipeFumero663 Recall@5 going from 0.75 to 1.0 with contextual plus hybrid and rerank is a real jump, though six questions is a small set to lean on. Curious how much of that came from the rerank alone. Are you growing the eval set as you add docs?
@jipneshjindal Money on the line plus photo-verify is a way harder promise than a streak counter. Curious how you judge the photo though, is anyone actually checking it or is it just proof you showed up? Feels like people will pick way too big a stake on day one.
@jawadhameedjhb@ProductHunt Free with no paywalls is a real differentiator here, most ATS tools score you then gate the actual fixes. On PH day, replying to every comment in the first few hours moves ranking more than the upvote ask does. Are you scoring against a generic model or per job description?
@KendallGridszy Offline-first is the right call for gear tracking, you're usually in a field or a rental counter with no signal when you actually need a serial number. Curious what the testers reach for daily, the receipts and insurance docs or just the inventory list?
@abhayk_01 The State pattern is the one place Rust makes you actually think, the classic version fights the borrow checker until you switch to enums. Trait objects vs generics is the same fork over and over. Is the blog post one from the book or your own spin on it?
@EcodeJR One-click retry is the right fix, digging through logs to re-run a failed transcription is the worst part of running this stuff. Curious if the retry reuses the stored audio or asks for the file again, and whether you cap retries so a bad file doesn't loop forever.
@thesnjvsharma A flight path as the focus timer is a nicer metaphor than another countdown, you actually want to land. Tracking flight history first is the right order too, the dashboard's only interesting once there's real data behind it. Does an early bail show up as a diverted flight?
@junaid_mohana The second one reads faster, "Where will your energy go?" gives you something to answer instead of a list to scan. The first feels more like a dashboard. I'd double check contrast on that yellow button though. Which one do people actually tap first?
@ChinesePowered "A focus timer with something to lose" is a way better hook than another pomodoro clock. Curious what actually happens to the capy if you bail mid-session, does it just look sad or do you lose real progress?
@s_munoz5 No signup and unlimited attempts is a bold combo, it either fills the board fast or one person just farms the crown all day. How are you stopping that, rate limit per IP or best score only?
@TonnyKamau3 170 new customers and $0 MRR is a pretty familiar shape for a focus app, people install it and never hit the paywall. Where'd those installs come from, Play Store search or your own posts?
@Axonris_ Cutting off after 3 attempts instead of 6 is the right kind of boring fix, most of that loop is the model retrying the same wrong idea with more words. Curious what happens to the ticket after a fast-fail, does it escalate to a bigger model or land back in the human queue?