Pinned Tweet
Agents run growth for my 6 mobile apps. I answer yes or no.
Since October: 17 experiments, 8 kept, 3 reverted, 6 running. Page conversion 19% → 41%. First debt added 47% → 56%.
Now doing the same for studios with 5+ apps. If growth eats your week and budget, DMs are open.
21
Felt like my AI factory had stalled. Asked Claude to find out why.
It hadn't. 134 PRs merged in 14 days. 36 more open. 24 Jira tickets.
The rest was waiting on me, spread across GitHub, Jira, Ops and old to-do files. No single list of what needs me.
11
Claude burned its 5-hour limit in 2 hours this morning.
One daily pass over the factory queue: store work, support, a review, a research doc. Then the sessions stopped.
Including the fix for exactly this problem.
4
1
42
First real number, from one task:
75 model calls, $9.44 at API prices. A third of it was the agent waking up after an hour and rereading the whole chat. $1 to $1.60 each time.
1
9
Since mid-October, ChatGPT has sent Money Goals 331 iPhone downloads. Google: 8. Those ChatGPT users buy Pro about twice as often.
So I asked ChatGPT 59 "which app should I use…" questions about my 6 apps.
We showed up in 0.
It reads developer websites first, not the App Store. Now rebuilding every app's page.
1
28
Quick intro
Solo founder. Finance apps. 10+ years, 1M downloads. Building an AI factory behind them.
Looking for people who ship: indie apps, agents, products with real users. Drop what you're working on.
3
51
About half of the people who downloaded Debt Payoff Box never added a single debt.
So 12 days ago I shipped a simple onboarding: 4 questions, and you see your debt-free date.
Since then 55% add a debt instead of 47%. And almost twice as many buy Pro.
Early days. Final numbers on Dec 23.
3
1
30
My AI support agent almost confirmed to a spammer that our email is alive.
Draft: "Please remove us from your mailing list."
Passed all 7 checks. All tested the reply. None tested the decision to respond.
Now a classifier proposes first: reply or close. Code routes. I decide.
1
18
Before Dues, I scored 14 finance app ideas.
Net worth scored highest. But its demand is on the web, not in the App Store.
Subscriptions: "cancel subscriptions" gets 10K-100K searches/mo. 3 apps in US Finance top-100 grossing.
Built that first.
2
39
My AI agent ran our app releases. Favorite line: "sorry, that's on me."
Builds with red local checks. Android when I needed iPhone. iOS versions never marked with tags.
500 free Codemagic minutes gone, +$60 in September, more in October.
Fix: bookkeeping is code now. Agents do creative part.
3
39
Update: I hit a wall. It's called "approved by Andrey". 17 tasks are waiting for me right now, across Ops, Jira, Todoist and agent chats.
I noticed two patterns:
- Simple UI tasks. iPad version of Money Goals: pure visuals. The agent sends screenshots, I comment, done.
- Complex logic tasks. Debt archive in Debt Payoff: calculations. I'm not shipping money math without checking the numbers myself.
The first was easy to approve. The second is still waiting on me: I need to test it on a real device and check the calculations.
What's next: splitting tasks by complexity and risk, so simple, low-risk ones ship automatically. Complex ones come with a full test plan: calculations, screens and flows, all in the PR.
1
41
Do I have domain addiction? 100+ domains bought since 2012.
Some dropped, a few sold, a handful in use, many still parked.
Every one of them was a "this is it" moment 😅
1
2
34
Is it worth automating app publishing?
I've been spending too much time on app publishing and ASO lately, and started noticing how much I actually do by hand.
For my new app Dues: Cancel Subscriptions I had to set it up in:
- App Store
- Google Play
- Firebase
- RevenueCat
Then prepare and upload all the ASO assets, policies etc. to App Store and Google Play.
Despite it being my 100th submission, I still spent a whole day on it.
I have 3 new apps to publish and 3 to resubmit in the coming weeks. Each step is small, and each can be done by an agent.
I'm starting to automate the whole thing now. Will share how it goes.
1
30
I am the best client for cloud and AI companies...
Started SFA Factory with the idea of building a fully sovereign Ops layer for my company. Great? Not so fast.
Ended up hosting Svelte + Postgres on Railway, slowly getting tangled in their quirks... Not too bad? Well...
Upgraded to Pro before getting a single piece of value. Why? SMTP is blocked on the Hobby plan.
Not Railway's fault, it's a great platform. It's me: I wanted sovereign and ended up buying convenience.
Again. Is there a way out?
Money Box 1.0 shipped in February 2015.
Money Goals 2.0 shipped today.
Back then: Objective-C on iOS, Java on Android.
Now: one Flutter app, rebuilt from scratch with Codex in February, finished with Opus 5.5 last week.
The topic never gets old: people dream, people save. 1M+ goals created in the app so far.
11 years between 1.0 and 2.0.
6
6
106
One of my biggest disasters: an app for angel investors, marketed with ASO.
Turned out nobody searches the App Store for that.
Lesson I keep relearning: check where your customers actually look before you build.
The app was cool though... domain name even cooler...
1
36
I have a dream: I'm working on something big and beautiful in full flow, while all the small and big tasks in my business are handled by a team.
I lived it before, leading big teams. Now I want to build the same with a team of agents.
Last week I worked synchronously with AI on one big topic: my new app, Dues. A few parallel Claude sessions, an immediate feedback loop, constant judgement.
It worked great: the work is done, the app is released, the quality bar is met. But the rest of the business (almost) stalled: none of the other projects moved.
My foundation isn't automated enough yet to let me create freely while the business keeps moving.
So that's what I'm building next: deterministic workflows and access rights, starting with simple processes (support, tasks) and moving to more complex ones.
One step closer to the dream.
1
56















