WHY USE MORE THAN ONE AGENT

LEARNING GOALS By the end of this chapter, you will be able to explain the difference between a single-agent system and a multi-agent system. You will understand why splitting work among specialized agents can help. You will understand a benefit called context isolation. Most importantly, you will learn when NOT to use multiple agents, because the honest answer is that most tasks do not need them. STARTING WITH AN EVERYDAY PICTURE Imagine you own a small bakery. When the bakery is tiny, you do everything yourself. You bake the bread, you serve the customers, you count the money, and you clean the floor. This works fine when there are only ten customers a day. Now imagine the bakery becomes popular, and a hundred customers show up every day, and they order custom wedding cakes. You cannot do everything well anymore. So you hire a baker, a cashier, and a cake decorator. Each person focuses on one job and gets better at it. The baker does not need to know how the cash register works, and the cashier does not need to know how to pipe icing. This everyday story is almost exactly the story of single-agent versus multi-agent systems. A single agent is the one-person bakery. A multi-agent system is the bakery with a team. And just like in the real bakery, adding people also adds new problems: they must talk to each other, they can disagree, and somebody has to manage them. That is why this whole Part exists.