DL eng building https://nitter.cf/t.co/o46R0w3gK4 - outfits from your closet. I write about b2c, sharing both wins and fails backed by @AntlerGlobal member @fdotinc @ns climber
SEA
Joined September 2012
- Tweets8.5K
- Following845
- Followers8.7K
- Likes13.6K
Pinned Tweet
WE BUILT ONBOARDING AROUND OUR BIGGEST WEAKNESS
nobody wants to photograph their clothes so we killed this step
aesty.ai builds your digital closet from your camera roll in ~2 mins
all on-device, nothing leaves your phone
design inspired by @lihachyov
live! 🚀
looks like “doing things that don’t scale” isn’t about doing totally unscalable things 😅
i think it’s about building a scalable product and doing parts of it manually until it’s automated - the line it took my friends Anna @burninganna and Alex @vargastartup a $100k/mo contract and two months of recovery to learn. good unfiltered read!
my halloween content pipeline this year:
- claude finds what's trending
- generates the looks with mcp.aesty.ai
- finds where to buy every piece with our “Shazam” feature
- i cut 2-3 seconds for hook from my build in public and other videos
🎃
day 230 building aesty.ai in public!
- doing some halloween specials :)
- still solving on-device/local run outfit making bug
- circling back to affiliate program
let's go!!
btw! this video was edited (including my make up) with @reelful by my friend @katedeyneka 🥰
day 229 building aesty.ai in public
- solving collision between promo codes and referral codes
- investigating on-device/local run misalignment for some algos
- continue influencers outreach + recording more FGC videos
let's go!
I didn’t buy costumes for my little demons, just used @reelful 🧛
@katedeyneka this is dangerously fun!! 😆
day 229 building aesty.ai in public
- solving collision between promo codes and referral codes
- investigating on-device/local run misalignment for some algos
- continue influencers outreach + recording more FGC videos
let's go!
day 228 building aesty.ai in public!
- pushing tech release
- polishing onboarding bottle necks i've found recently
- posted an article about my UGC/FGC pipeline, gonna share in comments
let's goo!
my DNA:
- shipping b2c apps with @im_moonko
- climbing & bouldering
- our rescued dog rocky
- bg3 and ghost of tsushima
- one piece, bleach and JJK
2 places modern ASR still breaks - accents and domain terms
most benchmarks often report clean native audio, so the gap is invisible until it hits your users
my friends at fluently built a bench for accents, which systems hold up on real non-native speech
Cool!!
Speech-to-text is solved. Unless you have an accent.
Over 1 million people from 180 countries have spent 580,000 hours speaking with Fluently AI tutors while learning a new language.
Nearly all of them have an accent, and transcribing their speech is still hard.
So we benchmarked streaming speech-to-text systems on spontaneous, accented English from our learners. Leaderboard and methodology in the thread.
English is the easy one. Our learners study 14 other languages too.
Working on tough speech problems? DMs are open.
already made one for slideshows
it’s fashion/lifestyle-heavy, but most formats work for any app
figma.com/board/ChAthIxdUSA9…
not a promo but this is a genius marketing strategy
10mo ago we did the same with covers.aesty.ai and got tons of downloads for aesty.ai and i still see our generated covers sometimes over there
i just open-sourced the results of my claude trendwatch: the UGC hook formats selling apps on tiktok right now
board + the skill to run it yourself 👇
figma.com/board/oy1cWG5xkzhf…
