Building the compound system — parallel review, the 50/50 rule, agent-native parity

The production practices that make the loop scale. Students learn three load-bearing
ideas. Parallelisation: the bottleneck is no longer human attention but compute,
so multiple agents and multiple features run at once — including many specialised
review agents (correctness, security, performance, testing, project-standards, and
more) examining a single change in parallel. The 50/50 rule: split engineering
time evenly between shipping features and improving the system that ships them,
because an hour spent building a review agent saves many hours of future review.
Agent-native parity: anything a human developer can see or do — run tests, read
production logs, open PRs, debug from screenshots — the agent must be able to do too,
or the human silently becomes the manual fallback. Connects forward to EB-7
(coordination) and back to sub-agents (EB-2).