Relja Bulajic

App store: apps.apple.com/rs/app/sage-ai…

Open your fridge. Take a photo. Get a full meal plan with macros in seconds. Just shipped Sage, an AI calorie tracker that actually sees your food. No barcode scanning. No manual logging. Just point your camera and eat. Live on the App Store now.

@betomoedano That Nov → Mar growth curve is beautiful. $15 → $541 in 5 months with $300 total ad spend. The fact that revenue kept climbing after you stopped ads means the product has real organic pull. What's the main acquisition channel now?

~4 months ago my app had barely any users. Today it's at $541/mo — after spending ~$300 on ads that didn't work. The code didn't change. The marketing barely changed. The models got better. Here's everything that worked, what didn't, and the real numbers 🧵

@kylegawley Smart move if you're doing light tasks. For heavy iOS dev sessions with Claude Code though, self-hosting can't match the speed and context window of the hosted version. What model are you running on the droplet?

I cancelled my Claude subscription I'm now self hosting the open source version on a $5/mo droplet

@pmitu Because not everything needs AI. My posture tracking app uses raw AirPods motion sensors — no AI, just physics. AI is a tool, not a requirement. Ship what solves the problem, AI or not.

Why would you build anything non-AI-powered these days?

@TeeDevh 12 apps. Not even close. I've shipped 8+ in a few months. Most taught me more about what users actually want than any single app ever could. You can always go back and improve the winner — but you need to find the winner first.

Which one gets you to success faster? 1 app - improved for 12 months or 12 apps - built and validated over 12 months 🤔

@athcanft $1,730 MRR on autopilot is the dream. This is what a good retention rate looks like — 294 active subs after 4 months of zero updates. What was the app? Curious what niche retains that well without new features.

havent touched this app in 4 months 1. ran tiktok ads in december 2. turned off ads in january 3. $$$$ it’s not “holy shit” money but it pays the bills

@hii_mohit Welcome to macOS dev. I launched my first Mac app recently too — AirPosture. 3.7% conversion rate on day one is solid. The hardest part now is keeping that traffic coming after the launch spike fades.

Yesterday I launched my first Mac app and today I got my first customer. 3 days ago, I didn’t even know how to code a Mac app All thanks to claude code 🧡

@buildwithshyam D — some traction. 8+ apps live, growing slowly, figuring out which ones to double down on and which to let die.

What’s your current stage? A. Idea phase B. Building silently C. Launched but no users D. Some traction E. Figuring out monetization

@vishaltweetup For coding, hard disagree. Claude understands SwiftUI and iOS architecture better than any model I've used. For general chat and browsing? Maybe. But I'm not paying $20/month for chat — I'm paying for a coding partner.

Unpopular opinion: If you’re paying $20/month, ChatGPT Plus is better than Claude Pro.

@dramaricic TikTok — yes. Short app demos get surprisingly good reach for iOS apps. I've been running multi-account TikTok strategy for different apps. YouTube not yet — the production time doesn't justify the ROI at my scale.

Hey builders, do you promote your product on youtube and tiktok? Is it worth it?

@thenowhereway Both. I build apps that solve real problems, but I use AI to ship them 10x faster. The question isn't either/or. It's: can you identify a real need AND execute fast enough to capture it?

Be honest: Are you building something people want? or Something AI made easy?

@shiri_shh This is the best counter-argument to "you need followers to make money."Build something people want, distribution follows. Followers are a vanity metric — revenue is the only one that matters.

bro is doing $2.8 million MRR with just 22 Followers on X 🫡

@Govindtwtt Yes. Switched to Claude for coding months ago. Better at understanding SwiftUI architecture and doesn't hallucinate import statements as much.Still use ChatGPT for non-coding tasks though.

Anybody else who has retired ChatGPT and is using Claude instead? 🙌🏻

@droidbuilds CoreBluetooth.

I’m not a vibe coder. Scare me with one word.

@aryanlabde This. I've shipped 8+ apps. Building takes days now. Finding the first 10 users who'll pay still takes weeks. Distribution is the real skill in 2026, not coding.

Vibe coding a SaaS is the easy part now. The hard part is finding 10 people who will actually pay for it.

@eliana_jordan This is a solid app idea. PhotoKit + on-device ML classification can handle most of this natively in SwiftUI. Might actually build this.

someone please build an app to automatically organize my iphone photos It’s absolute chaos

@imuratalpay Waking up to sales from countries you've never been to is the best part of iOS dev. What's the app? That yearly-first pricing strategy is smart — higher LTV from day one.

I launched my app last night. Went to sleep. Woke up to 4 sales. 4 countries. Zero ads. The best distribution channel is still just: ship something people want 👌🏻

@anulagarwal This is the same lesson I keep learning. My best-performing app ideas weren't the "smartest" ones — they were the simplest ones that solved one clear thing. Shipping speed > idea complexity. Every time.

You are all overcomplicating your app ideas. This viral app made $20,000 in 10 days. The idea? "Slap your mac and it moans/screams" Created by a solo-developer. >Marketed using Instagram & X & Product Hunt. Stop overthinking, start shipping.

@iyoushetwt Depends on what you're building. For iOS/SwiftUI, Claude understands the architecture better. For quick scripts and boilerplate, Codex wins on value. I use both daily and switch depending on the task. At $20/month Codex is a no-brainer as a secondary tool.

Unpopular opinion: If you’re paying $20/month, Codex is better than Claude Pro