>be me >2016 >working as taxi driver >bought first btc on 2 bucks to join redex pyramid >invited friends, didn’t withdraw in time rugged >bought eth at 10, sold at 11, thought im genius (idiot) >making pepe memes on xcp chain 2017 >trading on poloniex >x100 on nem >got hacked, lost 10 $btc 2018 >had 100k portfolio >didn’t sell cause waiting for million >watched it go to zero in one week >went to construction business >started hiring people, doing big projects 2021 >came back to crypto >nft boom >listed my pepe nfts on opensea >BOOM, made 300k for the first time >proof opensea activity in comments >caught luna crash lost 50k 2022 >joined stepn >printing hard >rugged again while on vacation in greece minus 40k >my kid was born best thing in life >started sybilling with a friend >made 100 accs farming sui arbitrum zksync starknet etc 2023 >arbitrum dropped 200k on all accs >reinvested >bought land to build house >expanded to 800 accs >zksync and others L2 rugged, closed the farm > $sui farming paid off got into ico with 70 accs ended up with 40k sui still holding still believe 2024 >started building house >tried youtube- failed >tried twitter, bored, quit >learned ai, started coding 2025 >finished house >broke up with girl, hard times >now writing bots and running acc farms >back on twitter again life goes in circles but im still here

built faceswap into sandwich for $1.64 the saas charging $30/month for this is still writing their PRD lol

asked sndwich to faceswap me into a sexy girl it worked in one go now you can be a sexy girl with sndwich in one click

everyone's bookmarking the same github repos nobody's actually using them these 5 are different open one and you'll wonder why you spent the last 6 months doing it the slow way 👇 1. photocraft photoshop, written from scratch in rust. no subscription, no cloud, no account. 56,586 stars in 13 days github.com/storytold/photocr… 2. filmcraft premiere, same team. they wrote their own h.264, hevc, prores and aac in rust every button is also a command, so an agent can cut, grade and export for you over mcp github.com/storytold/filmcra… 3. magpie every agent's model in one place. codex on deepseek, claude code on kimi, from the menu bar stop paying for whatever model your tool happened to ship with. 11,355 stars. github.com/yetone/magpie 4. strata qwen3.8-flash-next on your own hardware. one click on windows or linux, openai and anthropic api on localhost. 23,499 stars in 19 days. github.com/Niko1221/Strata 5. jev-ultrafast the fastest and cheapest web agent, from the browser-use team. 22,809 stars. github.com/browser-use/jev-u… every one of these was created in the last four weeks. the slow way was never the only way. it was just the one you had open.

Instagram would not let me promote a Reel it had just published. The video was vertical. The editor said 9:16. The file was 1079x1919. Nine sixteenths is 0.5625. Exactly. 1080x1920 is 0.5625; 1079x1919 is 0.56227. Nobody can see the difference. The ads check is not looking, it is comparing numbers, and it has no tolerance wide enough for one pixel. Where the pixel comes from: a browser editor records the video off a canvas, and the canvas is the board times the view scale. Before recording, the scale is set to one by multiplying it by its own reciprocal. In arithmetic that is one. In binary floating point it is 0.9999999999999998. The canvas is asked for a width of 1079.9999999999998, canvas.width is an integer, the fraction is dropped. We replayed that arithmetic across every window height from 300 to 2000 pixels. On 257 of 1701 heights, a 1080x1920 board recorded as 1079x1919. Not a random glitch. It depends on how tall your browser window happens to be when you export. Hence the lottery. Publishing and promoting check different things: publishing takes almost any shape, promoting demands an exact 9:16. That is why the Reel is already live, already collecting views, and still cannot be boosted. How to check your own file in ten seconds, and how to fix it: sndwich.app/blog/reels-aspec…

faceswap is now inside the editor no third party tools no exporting, re-uploading, switching tabs just drop the clip and swap one window. done

Looked up four AI accounts that actually sell something. Opened their profiles, wrote down what is visible, then checked our own invoices against it. The big one: the model is not the bottleneck, the volume is. 1,173 posts behind 1.1M followers. 1,500 behind 2.2M. An AI character removes the need to look, travel and film yourself. It does not remove the need to post every day. What it costs, from our own ledger: $0.0096 for a post rewritten in a persona's voice. $0.05 to $0.45 per second of video, depending on whether you generate the person or replace them. And the part nobody writes down: our month came in at $18.26 against the $2.00 the dashboard showed. Four profiles, the numbers behind them, and what breaks first: sndwich.app/blog/how-to-crea…

HashHedge is running a 48h Uptober deal 25% off $10K / $25K / $50K challenges stacks with HH Coins phase 1 target cut to 5% 10% cashback on any size code: UPTOBER25 ends 15 Oct, 23:59 Dubai hashhedge.com/?fpr=satty

i post every day and it takes me maybe 3 minutes here's exactly how inside sndwich: raw idea → hit generate → push to calendar one button no tabs. no copy paste. no "i'll do it later" the bottleneck was never the ideas

swapped a face in a video for literal cents took 99 seconds no editor. no after effects. no $200 freelancer dm the funny part is it just worked

tested 3 ai tools back to back this week → cursor: solid for editing, loses context fast → lovable: shipped a full ui in 40 mins → claude: still the one i trust for logic and copy

i asked sndwich to make a faceswap in video it took 4 mins and 0.78$ after that with 1 click you can post it across all your socials not a demo. actual shipped output

Already cashed out profits from HashHedge twice And now they're running a promo that's honestly hard to ignore: — Before, you needed 8% to pass the challenge — Now it's just 5% — Less pressure, same funding — I've already passed and cashed out twice at 8% — At 5%? That's a whole different game Grabbing a $100K account with the profits I've already withdrawn. If you're still thinking "maybe I should give it a shot"- this is your moment 👀

being a solo dev right now feels exactly like holding BTC in 2016 everyone thinks you're crazy the upside isn't obvious yet the tools are rough but they're working and the people who don't see it won't see it until it's too late i've been here before this time i'm not selling early

7 months building sndwich. here's what it actually taught me: → your first version of the product is always wrong → "almost ready" will eat you alive if you believe it → you'll rebuild the same feature 3 times before it clicks → distribution is harder than the build. by a lot. → the first launch is always quieter than you expected → month 4 is when most people quit → the product teaches you what it wants to be, not the other way around i almost stopped at month 4 glad i didn't

i ignored SEO for 6 months figured posting on X was enough then i checked where B2B buyers actually find products spoiler: not Twitter lol so now i'm doing the boring stuff: → keyword research around real buyer intent → pages built specifically around those searches → internal linking so Google follows the trail paid ads stop the second your card does SEO compounds build it once, keeps working

i stopped reading growth threads and my numbers went up the funny part is i don't think that's a coincidence less consuming, more shipping turns out the people writing "how to grow on X" aren't growing on X