A few weeks ago, my wife and I were running an errand, doing what every couple does, complaining about gas prices and the state of the world.
That’s when I said it: “If people knew how many years they had left from the day they were born, and were reminded every day how many hours remained on this planet, the world would be a better place.”
She paused. Then: “Great idea. Why don’t you build an app?”
My response: “Hmm. We could call it KarmaClock. Show people how much time they have based on their date of birth.”
That was the seed.
The thought wasn’t really about an app. It was about a human problem: we avoid thinking about mortality, and that avoidance is what enables us to be selfish with our time. A daily glance at a life-hourglass—341,566 hours remaining, 51.3% elapsed—turns abstract finitude into a felt thing. Felt finitude prompts a small action. Small actions, repeated, change behavior.
Today, KarmaClock is a free React Native app. iOS submitted to App Store, pending review. Android live on internal track. Built solo across ~30 hours of focused work over two weeks of evenings and weekends.
While running a 150+ person engineering org and a $200M P&L.
Here's what's actually under the hood:
→ 75 curated actions—small daily nudges toward kindness, health, and self-awareness across Physical, Mental, Social, Spiritual, and Nutritional categories
→ 211 passing tests, ~90% coverage (Jest unit + RNTL component + Detox E2E + k6 performance)
→ A 4-lane CI/CD matrix: PR checks → main merge integration → nightly E2E → release-tag full suite
→ Supabase backend with Row Level Security on every table, served via a CDN-cached edge function
→ Three custom Claude Code skills I wrote along the way: /release, /spec-audit, /pm-review
→ Hosted privacy, support, and account-deletion compliance pages
→ 31 commits, 3 released versions, 4 distinct phases—exploration, scaffolding, first launch, hotfix, second platform submission. The full software lifecycle, not a prototype.
Here's the part that should matter to every CTO and engineering leader reading this:
A scoped, production-deployed mobile app like this—iOS and Android, with auth, RLS-secured backend, CI/CD, full test coverage, store-compliant privacy and account-deletion pages—is not a one-engineer project in a healthy SaaS shop. It’s a coordinated 6 to 8 calendar weeks across 2 devs, a QA, partial DevOps, partial IT/SecOps, partial PM. Realistically ~1,000 person-hours.
I did it in ~30 hours. Solo. On week nights.
That’s roughly 5x faster on calendar time. ~30x fewer person-hours.
Four things I learned that I’m bringing back to my org:
1. Agentic AI doesn’t replace engineering discipline. It amplifies it.
I went spec-first. Functional and non-functional requirements before any code. Test specs mapped to spec IDs. 85% coverage gates in CI. The discipline I’d demand from my team, I demanded from myself—and from Claude as my pair programmer. The result wasn’t faster garbage. It was 211 passing tests at ~90% coverage, shipped in 30 hours.
2. The bottleneck moved.
For 27 years, the bottleneck in software was typing—translating intent into code. With agentic AI, the bottleneck is clarity of intent. Engineers who can write a tight spec, decompose a problem, and exercise judgment on architectural tradeoffs are now dramatically more valuable. Engineers who can only execute against a ticket are not.
3. Hands-on still matters at the exec level. Maybe more than ever.
I can’t credibly lead AI transformation across my org if I haven’t felt it in my own hands. KarmaClock was a forcing function: stay current, stay technical, stay honest about what works and what doesn’t. Every leader rolling out Copilot or Claude to their teams should ship something themselves first. On a week night.
4. The economics of software are being rewritten in real time.
5x faster on calendar time. ~30x fewer person-hours. That’s the number every CEO and board should be wrestling with right now. If a production-quality mobile app shrinks from a 6 to 8 week multi-engineer effort to a solo 30-hour evening project, what happens to your roadmap? Your headcount plan? Your competitive moat? Most orgs haven’t internalized this yet. The ones that do first will eat the ones that don’t.
But honestly? The economics are not why I built it.
I built it because the thought in the car—we treat time like it’s infinite, and it isn’t—was worth turning into something real. And for the first time in my career, the gap between having a meaningful thought and shipping a meaningful product was small enough to cross in evenings, alone, after the kids went to bed.
That’s the part I want every technology leader to feel.
It’s not that AI builds apps faster. It’s that AI lets you ship ideas at the urgency of the idea itself.
KarmaClock launches publicly soon. Comment “clock” and I’ll send you the link when it’s live on both stores.
Make every hour count.
#AgenticAI #EngineeringLeadership #SpecDrivenDevelopment #BehaviorChange