I build Mac/iOS apps and try to turn them into sustainable businesses
Sharing the real process: product decisions, numbers, what works, and what doesn't :)
Marquee 1.1 is out โ and it's the biggest update I've shipped.
Marquee turns your raw app screenshots into a full Mac App Store listing: designed galleries, preview videos, copy, keywords. Version 1.1 makes it a studio that thinks with you.
๐ค Your coding agent can now drive the studio. Marquee ships with marquee-mcp โ a free MCP server that lets Claude Code or Codex build a complete listing from one prompt: real screens in, store-exact files out. Every agent edit is recorded and traceable. Publishing stays human.
๐ ASO Autopilot. Point Marquee at your app and it reads your screens, works out what you're building, then researches your actual market โ live storefront searches, competitor verdicts, and a keyword plan where every claim has a receipt. The new Research Room keeps the evidence: coverage, audit health, and a copilot that cites its sources.
๐ฌ AI in the room. An on-device chat panel that doesn't just answer โ it does. "Make it warmer." "Match the background to the app." Every change lands in History, โZ undoes it. Nothing leaves your Mac.
๐ My Apps. Your launches, tracked honestly โ daily downloads straight from your own sales reports, every storefront.
The video below is the real workflow, recorded โ not staged. An agent builds a listing for my other app, Spaceback, and the studio takes it from there.
Free on the Mac App Store โ https://t.co/finaT7LIVe
*still looking for app store indie devs to test the app and give feedback, comment "Marquee" or DM me and I'll provide a code that unlocks the 50$ Pro subscription for free (lifetime)
#macOS #indiedev #buildinpublic #ASO #AI #ClaudeCode #MCP
Slowly but surely, Spaceback for iOS is getting there
13 active plans now
Nothing crazy. No viral launch. No influencer campaign. No TikTok machine behind it
Right now, the only distribution channel Iโm actively using is Apple Ads
For Halloween, Spaceback will become Spellbook. ๐
Iโm already working on Spacebackโs Halloween In-App Event, and this is the actual UI direction Iโm shipping.
Would you use this?๐ป
Iโll be sharing the UI as it slowly mutates into Spellbook over the next few weeks. ๐
Had a really valuable conversation with @geckodotdev about Forgot AI today ๐
One thing stood out to me:
The best product feedback isn't โthis is cool.โ
It's โhere's exactly how I would use this.โ
Vlada gave me a very specific use case: saving TikTok/YouTube videos and having AI understand the important parts so they can be found and used later.
We also went deep on retrieval, organization, natural-language search, connecting saved context with other AI tools, UX, user behavior, and distribution.
My biggest takeaway:
Don't keep adding features just because you can.
Make the core experience great, put it in front of people, and watch what they actually do.
That's what I'm focusing on next with Forgot AI.
Appreciate the time and honest feedback, Vlada ๐
Back to building. ๐
Shallow posts sell.
Iโm guilty of this too
Post about a simple app doing $100k MRR? Tons of attention.
Show a polished UI? Same thing.
Write something genuinely in-depth about distribution, retention, pricing, operations, or why a product actually works?
Buried๐
I pay Apple $99/year to be part of the Apple Developer Program.
Apple takes 15โ30% of what developers make.
And when thereโs a problem with an App Store payment, there is no Apple Developer Support number you can just call.
You submit a form.
Then you wait.
For a business-critical issue involving money.
@Apple , developers arenโt just users who forgot their iCloud password. Weโre businesses running on your platform.
Give us a phone number.
@AppleSupport
A quick look at the workflow:
1. Mapped onboarding as six explicit states: promise โ goal โ format โ source โ demo โ paywall. Each screen gets one message and one decision.
2. Used https://t.co/PjXSWHozTe via MCP to bring complete onboarding and paywall flows into context, then compared sequencing, copy density, trial framing, and free paths.
3. Turned those patterns into constraints: demonstrate the product before pricing, keep the free option visible, and cite the exact source behind generated content.
4. Created the supporting images with Higgsfield and assembled the final motion in Fable 5.1.
5. Tested the whole sequenceโnot just individual screensโfor pacing, back navigation, tap targets, Dynamic Type, and VoiceOver.
Takeaway: fewer screens isnโt the goal. Less time before the user sees proof is.
@shownotover@uriel_builds high intent audience is expensive. tiktok is in your budget + it compounds. find a winner format and dump the whole $100 in one video. worst case scenario it grows your account
Another reason to stop overthinking your app launch:
Even your App Store rating isnโt permanent.
If v1 disappoints, fix the product, ship an update, and you can reset the summary rating for the new version.
The written reviews stay, but youโre not stuck carrying a bad launch rating forever.
Ship earlier.
source: https://t.co/8U0F06LZeX
Curious how people are actually using local models in their day-to-day workflow.
Especially now that Qwen 3.8 is out.
What do you run locally that you wouldnโt just use Claude/ChatGPT for?
Coding? Private data? Agents? Batch processing? Something else?
Would love to hear actual workflows, not benchmarks
Consumers donโt care if your app looks vibecoded.
They care if it solves their problem.
Polish matters when the product works. Before that, a beautiful app that does nothing useful is still useless.
https://t.co/48q7bO9fwe