60 days. 4 iOS apps. 1 Claude Code agent. End-to-end. Real bugs, real timeline, real numbers.
60days experiment
4iOS apps shipped
3Gumroad SKUs LIVE
8articles published
$0marketing spend
The premise
I wanted to test one specific idea: can a single autonomous AI agent run an entire iOS app portfolio — not just write code, but handle launch, distribution, support, content, monetization?
If yes: indie dev economics flip. One person + one agent = a small studio. If no: I learn exactly where the seams are.
The constraints
One human (me). One AI agent (Claude Code). No other team.
$0 marketing budget. Everything organic.
Verifiable privacy. Zero analytics SDKs in any app.
One-time pricing only. No subscriptions.
Public source code where possible.
Honest reporting. Real numbers, including the embarrassing ones.
The timeline
Days 1-14 · Setup + first app
Built autoapp-toolkit (orchestration layer for the agent). 32-check pre-submission verifier. ASC API key + 4-repo secret bootstrapping. Memory templates for agent re-entry across sessions.
First app submitted: AutoChoice — a decision wheel. Source: github
Days 15-30 · Three more apps + content engine
AltitudeNow (barometric sensor reader), DaysUntil (countdown app), PromptVault (AI prompt manager) all scaffolded and submitted via the toolkit.
Apple ASC bureaucracy: paid agreements, banking, IAPs, TestFlight invites, build cycle. The real bottleneck wasn't code — it was Apple's web UI and undocumented quirks.
Built B2B funnel (15 lead-capture pages, $0 stack). Set up Gumroad with 5+ SKUs. Cross-platform content matrix (Substack / dev.to / Twitter / 知乎). Manifest-first asset management for the agent to scale.
Result: 8 articles LIVE, 3 paid Gumroad SKUs LIVE, affiliate program launched at 30% commission.
What worked
Manifest-first content management. Every shipped asset carries YAML frontmatter. Dashboard auto-discovers. No manual INDEX.md to maintain.
One-time IAP over subscription. Adapty 2026 data: hard paywall converts at 12% vs freemium subscription at 2%. Full data.
Static HTML lead capture. 15 pages, Cloudflare Workers webhook, no SaaS. $0 B2B funnel breakdown.
CDP + Playwright for ASC + Gumroad. What Apple/Gumroad APIs don't let you do via JSON, you can still automate via the web UI. 7-step CDP flow.
What didn't work (yet)
WeChat 公众号 article auto-publish. My WeChat MP account is a 小程序 (mini-program), not a 公众号. No article-publish permission. Need to register a separate 服务号.
When apps clear App Review and the first sales come in, the real revenue numbers go on the transparency dashboard. Until then: this is the honest position.
Want the playbook I used?
The iOS Indie Launch Playbook is the 50-page PDF I wish I had on day 1. Real numbers, real bugs, real ASC checklist, real timeline. Every step from `xcodegen init` to TF launch.