My iOS app is 4 months old and does $944 MRR. This is everything I use to run it.

@GainFrameApp
АНГЛИЙСКИЙ10 июл. 2026 г.
389K
10
0
4
15

Суть

A solo developer shares the comprehensive 'org chart' of AI agents and automated tools used to scale an iOS app to $944 MRR, focusing on SEO, content marketing, and lean engineering.

My iOS app has an engineering team, a release manager, a content studio, an SEO department, and a data analyst.

It's one person and a pile of Claude Code skills.

GainFrame is 4 months old. $𝟵𝟰𝟰 𝗠𝗥𝗥, 𝘂𝗽 𝟰𝟱% 𝗶𝗻 𝘁𝗵𝗲 𝗹𝗮𝘀𝘁 𝗺𝗼𝗻𝘁𝗵, 203 subs. This is the whole org chart, with the actual workflows:

GainFrame - inline image
GainFrame - inline image
GainFrame - inline image
GainFrame - inline image

𝗘𝗡𝗚𝗜𝗡𝗘𝗘𝗥𝗜𝗡𝗚

SwiftUI + SwiftData. XcodeGen generates the project from YAML, so the .xcodeproj isn't even in git.

Claude Code writes most of the code. A CLAUDE.md holds the architecture rules (navigation patterns, design system, what never to touch) so every session writes code that matches the app.

QA is an agent: it builds, installs in the simulator, and drives the feature through RocketSim's CLI. Reads the accessibility tree, taps by element ID, screenshots each step. A green build says it compiles. This says it works.

Backend: Supabase, 30 edge functions. The AI coach streams Gemini 3.5 Flash and calls 17 tools against real user data: photo scores, Hevy volume by muscle, Strava activities, conversation memory.

AI cost guardrails, because the coach runs ~$257/mo: Slack alert past $35/day total or $5/day per user, and CI fails any PR that grows the coach's system prompt.

𝗥𝗘𝗟𝗘𝗔𝗦𝗘 𝗠𝗔𝗡𝗔𝗚𝗘𝗥

One command. Checks the live App Store build to detect new-version vs follow-up, drafts release notes from git log, bumps, tags, archives with fastlane match, uploads to TestFlight, syncs metadata in 9 languages. Screenshots auto-mirror en-US → UK/AU, es-ES → MX.

Work tracker: GitHub Issues with P0–P3 labels. One command picks the top issue and starts working it.

𝗖𝗢𝗡𝗧𝗘𝗡𝗧 𝗦𝗧𝗨𝗗𝗜𝗢

A TikTok comic factory. Gemini draws ONLY the mascot art (base64 reference images keep it on-model). Python/Pillow stamps every word: fixed font, fixed positions, fixed red. 4 cents a slide, 150+ carousels.

Why the split: letting the model render text made every slide drift. Art is AI. Typography is code.

What performs: myth-bust hooks ("WILL CARDIO KILL YOUR GAINS?") start comment wars, and comments are the algorithm. The app appears in 1 of 3 carousels, final slide only. More and reach dies.

Second pipeline: "top 5 fitness apps" listicles from real iTunes API assets, zero AI images, my app planted #3 between Strava and Hevy. Outperforms putting myself first.

Plus a $49 promo video from a Reddit cold DM that beat my own screen recording.

𝗦𝗘𝗢 𝗗𝗘𝗣𝗔𝗥𝗧𝗠𝗘𝗡𝗧

141 posts + 11 free calculators on Cloudflare Pages. Hosting: $0. Keyword tools: $0.

The pipeline: Google autocomplete → live SERP reads → my own Search Console over MCP → scored backlog → generator drafts → I edit → git push deploys.

7 clicks a day → 100 in 90 days. Best page is a "best AI body fat apps" roundup at ~170 clicks/week.

What I learned: shopping-intent posts (roundups, X-vs-Y) get clicks. Informational posts die in the AI overview. Free calculators rank faster than articles and pull the backlinks articles never get.

𝗥𝗘𝗩𝗘𝗡𝗨𝗘 𝗧𝗘𝗔𝗠

RevenueCat. Yearly has a 7-day trial, monthly has none.

Live paywall A/B: leading with "$3.33/mo billed yearly" converts 𝟯𝟬.𝟭% of views into trials vs 24.1% for the version explaining the trial timeline. Same price, different frame.

Every Pro feature gates with a preview of itself: your real muscle map, breakdown blurred. Show the thing, lock the depth.

Webhook → Slack on every purchase, cancellation, billing issue, with a "find in PostHog" link. Trial cancels arrive flagged "reach out" and I email the same day and ask why.

Trial→paid: 𝟮𝟭% → 𝟰𝟰% in 5 weeks. $0 ads.

𝗗𝗔𝗧𝗔 𝗔𝗡𝗔𝗟𝗬𝗦𝗧

PostHog + GA4 + Search Console wired into Claude over MCP. "Why did installs dip Tuesday" is a question I type, not a dashboard safari.

Surveys write to my own Supabase table. The vendor's survey UI was silently capturing 47% of responses.

Every raw data pull is committed to git in dated folders, so any past decision can be re-audited against the data that drove it.

The fix that mattered most: my analytics showed 10 trial conversions where RevenueCat had 48. Fixed measurement before optimizing anything.

𝗧𝗛𝗘 𝗗𝗘𝗣𝗔𝗥𝗧𝗠𝗘𝗡𝗧 𝗜 𝗙𝗜𝗥𝗘𝗗

Paid ads. $5,674 in 3 months: Apple Search Ads $2,498 (443 installs, $5.64 CPI), TikTok $1,881, Reddit $1,295 with ~zero tracked conversions.

Best case: $𝟭𝟭𝟰 to acquire a payer with an $𝟭𝟴 LTV. Killed it May 10. Growth came back on channels that don't stop working when the card does.

𝗪𝗛𝗔𝗧 𝗜𝗧 𝗔𝗟𝗟 𝗖𝗢𝗦𝗧𝗦

Gemini ~$260/mo. Hosting $0. SEO tools $0. Slides 4¢ each. RevenueCat and PostHog free at this scale. Apple $99/yr. Plus about two focused hours a day.

If you're starting one of these, steal in this order: webhook → Slack first. Fix your tracking second. Free calculators third. Comparison posts fourth. CLAUDE.md fifth.

Full writeup with screenshots of everything is on the blog, link in the reply.

Переделать в YouMind

Превратите одну вирусную статью в полноценный рабочий процесс создания контента

Собирайте источники, расшифровывайте паттерны, создавайте активы, пишите черновики и публикуйте контент из одного рабочего пространства ИИ.

Исследовать YouMind
Для авторов

Превратите ваш Markdown в аккуратную статью для 𝕏

Когда вы публикуете длинные тексты, изображения, таблицы и блоки кода, форматирование в 𝕏 становится мучением. YouMind превращает полный черновик в Markdown в чистую статью, готовую к публикации в 𝕏.

Попробовать Markdown для 𝕏

Другие паттерны для анализа

Недавние виральные статьи

Смотреть другие виральные статьи