Single Agent versus Multi-Agent: The Core Tradeoff

Start with the cleanest possible definition, the one used across these articles: an AI agent is a system that uses an LLM as a reasoning engine to decide the control flow of an application. From that base, there are two ways to organize the work.

A single-agent system is one focused worker handling a task from start to finish. It keeps one continuous thread of thought and action, so every step has access to everything that came before it.

A multi-agent system is structured like a team. A "lead" agent breaks a goal into smaller subtasks and delegates them to multiple "worker" agents, often running at the same time, then combines their results.

Most of this chapter is about understanding the precise tradeoffs between these two, because the rest of the debate (Chapters 2 through 5) is really an argument about when each one is the right choice.