The model was never the hard part.
One engineer and one designer ship and operate SocialScoreKeeper LLC's app on iOS and Android. What made that possible is the structure built around the agent since Claude Code launched in February 2025: the standards it reads, the gates it cannot talk its way past, and the memory that stops it relearning the codebase every morning.
What is actually running
Four layers. None of them are novel on their own. The work was deciding where each one has to be strict and where being strict just slows everything down.
The fleet
Five specialized roles rather than one general agent: explore the codebase, write the spec, plan the work, implement it, then inspect what came back. Each role gets its own instructions and its own tools. Splitting them was the change that made long jobs survivable. A single agent asked to do all five drifts by the third one.
The gates
The checks include a custom static analyzer Steven built for his state-management rules and published to PyPI. Twenty-two standards documents and a project constitution set the rules the agents read before they touch anything. A hook blocks the session if the work tries to finish without a self-review.
The tools
Dozens of command-line harnesses wrapping the services we actually depend on: Supabase, Stripe, RevenueCat, App Store Connect, Google Play, Cloudflare, Sentry. An agent operates the real system instead of describing what it would do. Each one is permissioned, logged, and safe to run twice.
The watch
Every agent session reports to a local OpenTelemetry collector with a live view of what is running, what it cost, and where it stalled.
Verification means trying to prove it wrong
An agent asked "is this correct?" will tell you yes. So findings get handed to independent reviewers whose instructions are to refute them, and a finding only survives if the attempt fails. When several reviewers are worth running, they get different lenses rather than the same one repeated: correctness, security, whether it reproduces at all.
That one change caught more real defects than any amount of prompt tuning did.
SocialScoreKeeper, our sibling company's consumer app across six sports on iOS and Android, extended and maintained by one engineer and one designer.
Privacy and compliance work covering COPPA, GDPR and the UK Children's Code, because the app holds photographs of children and that is not a thing to improvise.
This was proven at two people, not two thousand.
We have not run this inside a large engineering organization, and we are not going to pretend the politics of one are the same problem. What we can say is that the parts that carry the weight are portable, because none of them depend on trust. A gate that a machine checks does not care how many teams are behind it, and a standard written down is a standard whether one person reads it or forty do.
The part that would not survive the move is the part we would want to talk about first: who inside the company owns it after we stop pushing.
Written out at length, free, no email wall.
The environment is the leverage is the full account of the method: the standards, the gates, the memory, and what each one was worth. About sixteen minutes.
Incentives are the architecture is the other half: why a system's structure decides its behavior more reliably than anyone's intentions do.
If you want the teaching version of this for your own team, that is DayLight AI. If you want it built, start here.