Saba

@mariosaputra Seems on pair with Opus 4.6 on the “CursorBench Score”

Using Composer 2 now. It seems on par with Opus 4.6 and also costs less, which is great since I’m not continuing my Claude subscription.

@alexolim_ How? A 20€/video from a brend new account, is not cheaper then a 2$ cpm on meta ads imo

high-volume UGC is the cheapest testing engine ever built for consumer apps skip it and go straight to paid, and you're choosing the most expensive way to learn the same lesson

@mattgittleson Thanks for adding this note g

an underrated part of this is that ernesto had filmed + made content himself before he automated it you can’t make good automated content before you even know what good content looks like get filming now.

@lucasmoscon__ Let’s gooo

This is what it looks like to live in a hacker house 24/7 in 2026 2 months in, 4 months remaining 🇦🇷🤍

@ErnestoSOFTWARE Bro i have tested with my app and this hook doesn’t work anymore

$500K/mo for an app that is free on your iPhone 😭 Their app is basically a clone of the alarm app on your phone. Except it won't turn off until you take a picture of something random This app works because it’s marketed like a pain point. Being late + and snoozing your alarm These apps all follow the same marketing playbook Unhinged caption + app demo And it works, this video alone got 4.7M views And probably 10k+ downloads If I had an app like this → I would hire 10 UGC creators → 30 videos/mo each → "POV: my alarm made me ___" → the funnier the task, the more views That’s it.

@da5ch0 @bcherny Hi! Where is image gen?

@bcherny gonna be real: didn’t want voice, didn’t want image gen. both are frankly very mid

Update: this is now rolled out to 100% of users

@ErnestoSOFTWARE 500k micro influencer? Link one of yours that accepted 1$ for 1000 views because i talked to ~300 (not dm, talked), and they all declined. What i am doing wrong?

Literally the only guide you need to build a $10k/mo app in 30 days ( source: battle scars )

@claudeai Cute

Claude can now build interactive charts and diagrams, directly in the chat. Available today in beta on all plans, including free. Try it out: claude.ai

@andrew_cfg Hi! What is the name of this text animation? Can’t explain to Claude code how to do that 😕

✅ Chart scrubbing. It went pretty smoothly, actually. I had to drop the horizontal line from the scope, though. It differs from chart to chart and will cause more confusion than clarity. Next are the filters 🎛️

@seraleev I don’t think that widgets can be generalised this way. There are very specific use cases where thay are very useful, but in most cases they are not. Imo they are useful is they can be shared/flexed to friends

I started thinking about building home screen widgets for my apps. But do people actually use widgets? Is it worth spending time on them? I dug into the data and the results were pretty interesting. Sharing a few screenshots 👇

@ErnestoSOFTWARE How to lose token and fill your context window:

Don’t waste your time.. this is literally the only prompt you need to clone any design Into your app.

@CalebDeannn I think that this is just a move to make an ad for arcads like many other X bros have done during these weeks. He actually rage baited you

This is a stupid take. What people are missing is UGC’s UVP isn’t only being able to create mass content and spam TikTok’s. It’s having the ability to manage 10, 20 creators posting simultaneously across multiple social facets. Yes, you might be able to reproduce AI UGC videos which are almost indistinguishable from regular UGC videos. But… with AI UGC are you able to post hundreds of pieces of content across multiple accounts on TikTok, Instagram, Facebook and Youtube? Whilst testing: - Video formats - Creator profiles - Hook types (visual and verbal) - Demo videos - CTA integrations - Managing video comment sections The real value in UGC lies in having the ability move at lightning speed, allowing fast iteration and to double down on what’s working, WHILST posting at scale This is simply not possible with AI UGC...

@simonecanciello Too easy to cheat

a $100k app idea: every page you read, you earn 1 minute of screen time.

@seraleev @RevenueCat 17 days trial? It’s an average or it is really 17 days?

Got the annual subscription report from @RevenueCat based on 115,000 apps. Some interesting takeaways: ASO-only apps are dying. The bottom segment dropped 33% YoY. Want growth? Be in the top 25% – they grew 80% YoY. Reality: ad budgets = growth. AI apps are still hot. +41% revenue per paying user. But retention is weak. Found a way to bring users back? Good! Just pour more ads in. A stat I personally don’t like: Hard paywalls convert 5× better than freemium (10.7% vs 2.1% in 35 days) with almost the same 1-year retention. And another question: Are 3-day trials outdated? Trials 17+ days convert 70% better (42.5% vs 25.5%).

@jogicodes Let’s fking go

After thinking about it so much, I quit my job. I got a plane ticket to Vietnam to pursue my dream of becoming a successful entrepreneur. Currently I'm living off savings, coding 12 hours a day, and I know one day it will be worth it. Wish me luck.

@ErnestoSOFTWARE Btw you don’t train anything, the best you can do is like give nanobana pro 2/3 reference images to gen the next character but real exercises are very hard to replicate bc involves hands and equipment

@ErnestoSOFTWARE “You train your openclaw on a consistent style”. Like how? Blue bro is not AI generated for shure, most of them are handwritten characters. I

This is crazy… AI character slideshow accounts Are the easiest way to scale your app To $10k/mo+ → You train your open claw agent on a consistent style of character → You have him learn the correct methodology for your niche → He generates 50 hooks and angles → You have him batch generate 50 slideshows with CTAs to your app → You automate sending the draft to your TikTok. All you have to do is click post And reap the benefits. Just look at how many of these accounts generate millions of views:

@alexcooldev The only "automation" here is postiz, the video gen need lots of checks and control that is impossible to automate, you always have to check if the video is good enough and re-make it if needed (10€/video bit expensive!). + you can do all this in cowork/cc, no need openclaw imo

@jackfriks @postbridge_ i think that your mcp crashed: Error: HTTP 405: Invalid OAuth error response: SyntaxError: JSON Parse error: Unexpected identifier "Method". Raw body: Method not allowed FUNCTION_INVOCATION_FAILED

only thing missing is adding a social media MCP like @postbridge_ so you can also get claude to manage all your accounts and post to them from one place...

@ghulamio Lessgoooooo bro, you are not alone 😬

@Sabatin02 Hey, I'm also building a B2C app while doing my masters in AI. Let's connect!

🎉 New app is live 🎉 That's the app I built for the #Shipyard2026 challenge. I know that there are lots of app like this, but the user experience and reliability of this one i think it's remarkable. You can manage your shopping list just with audio. apps.apple.com/us/app/toque-r…