Five ideas
The book makes five arguments about one thing: leading a population of AI agents through the knowledge core. One chapter each. The first is what the population is made of: individuals, imperfect like us, each bound to one person who answers for it. Everything that follows is how to run an organization of them. The second is the problem.