Hey, Iβm Savelii Stepura, 30 y.o. indie dev from Georgia
Today, Nov 1st, Iβm officially starting my journey to $10K MRR in solo mobile development.
Currently at $1K MRR β Letβs build something meaningful.
#buildinpublic#indiedev#solopreneur
Still grinning about this: my best rated app out of 15 is a Bible study app. 4.8 stars, 225 ratings, zero ads, zero launch tweet.
People found it because they were already looking for it. That's the whole trick I spent years missing.
New model from Google πΊπΈ: Gemini 4 Carbon, now live on DesignArena and in Google AI. It ranks #1 on the Artificial Analysis Intelligence Index v5, and the arena run includes mobile and native app modes. Watching and testing π
"Just raise your prices."
Sure, but a cheap client doesn't become a good client at $2k. They become a loud one.
I price per stage now. Not because it earns more, because it ends the argument about what's in scope.
Claude Fable 5.2 just landed on DesignArena and the Anthropic API πΊπΈ. Built for long agentic coding runs, $8 in / $40 out per million tokens, down from $10 / $50. On my test list π
Meta's AI now answers from a watch. Talking to my wrist felt silly in 2015, now it's just how I check things. watchOS apps are about to get a lot more interesting.
Submitted at 11pm, convinced review would take a week. Approved in 4 hours.
Next build, one string change, three days and a rejection. Apple's review time is just weather. I check it, I don't plan around it.
AI writes the first 80% fast. The last 20% is where my apps actually live.
Memory leaks, offline state, a rejection over one Info.plist string. Nobody vibecodes their way through App Store review at 2am.
Instagram now guesses your age from behavior signals, worldwide. Creepy and impressive at once. The part I keep thinking about: on-device age signals could become a standard iOS primitive, and that changes how we all build kid-safe apps.
AI gave me 15 apps' worth of code in a weekend and zero apps' worth of users.
Turns out the App Store doesn't have a "generated really fast" ranking factor.
AI makes writing code cheap. It doesn't make App Store review faster.
My weather app still waited 18 hours after a rejection. Screenshots, privacy strings, support page, refunds. That's where solo time actually goes, and no model is touching it.
A coffee machine that reads your beans and adjusts the grind on its own. Honestly that's a nice bit of engineering π Feels like the sensor-plus-logic trick is leaking into every object in my kitchen now.
Being a solo dev means my standup is me telling myself the weather app is fine, and me disagreeing.
Guess who files the Info.plist rejection ticket. Also guess who closes it.
New on the Arena: gemini-omni-1.2-flash from Google πΊπΈ. Video generation and editing, clips up to 15 seconds (up from 10) and 30% cheaper than Omni 1.1. On my test list this week π
Scouts worked because 2012 had fewer people with money and opinions. Now every angel syndicate does the same job for free. Sequoia didn't lose faith in the model, the model stopped being scarce.
15 apps on the App Store, all shipped from one laptop. Still gets me that one person can design, build, submit and support a product users rate 4.8. No standups. Just Xcode and a to-do list.
Founders ask me what features will make their app take off.
Across 15 apps on the App Store, the ones that did anything had a reason to be opened without me reminding anyone. That's it.
Features were never the variable. Reason to return was.
New image models from OpenAI πΊπΈ: GPT Image 2.6 Flare and Sunburst. Both live in the API and already on DesignArena across logo, graphic design and image to image modes. Watching and testing π