subagents are an org chart you can rewrite at runtime.

managing six agents in parallel is managing six junior engineers — except you can change their reporting line mid-task.

in human teams, the org chart is the slow-moving infrastructure. you change reporting lines once a quarter, and even then it's painful.

with subagents, the org chart is a parameter. i decide at runtime which agent reports to which, who consolidates, who reviews. and i can change it mid-task without any of the political fallout.

this is the underrated power of the pattern. it isn't 'agents do more work in parallel'. it's the structure of the work is now plastic.