Three files have to be reworked in a fixed order, because each change rests on the one before it. A colleague proposes standing up a team of five teammates to get through it faster. What is the better call here, and why?

The principle — a team pays off when the pieces can run without waiting on each other. When they cannot, its coordination is a cost with nothing on the other side.

Why the key is correct — Anthropic names this exact case: for sequential tasks, same-file edits, or work with many dependencies, a single session or subagents are more effective. The scenario hits all three at once — one order, three shared files, every change resting on the last. The cost is documented as well: agent teams add coordination overhead and use significantly more tokens than a single session, and that is charged whether or not the work can use the arrangement. The line between the two shapes is what the work needs rather than how much of it there is. Use subagents when you need quick, focused workers that report back. Use agent teams when teammates need to share findings, challenge each other, and coordinate on their own. A fixed sequence needs neither sharing nor challenging; it needs step two to start after step one lands.

Why the others are wrong — the first reads size as the trigger, when dependency is. The second buys headcount, against advice that three focused teammates often outperform five scattered ones. The third treats the two arrangements as interchangeable, which is the one thing the comparison in the documentation exists to deny.

Remember this — ask whether the pieces can proceed independently. If they cannot, a team only adds a bill.

Sources — Anthropic's Claude Code documentation on orchestrating teams of sessions.