Solo production development with AI agents: enforce rules, don't explain them

Short summary

A founder recounts rebuilding an entire project solo using AI agents by enforcing discipline through machine-held rules — pre-commit hooks, CI gates, and branch protection — rather than relying on agent memory or long contexts. In 18 days with 655 commits, they shipped a full vertical stack including backend, web, mobile, CI, and deployment. The core insight: when rules are enforced by infrastructure, agent discipline is free, and fresh sessions become a feature rather than a limitation.

• •Agent discipline should live in repository infrastructure (hooks, CI, branch protection), not in model context
• •Fresh cheap sessions outperform long context windows because rules are enforced by machines, not instructions
• •18 days solo with 655 commits produced a working production system where a team's first attempt left no reusable artifacts

Generated with AI, which can make mistakes.