Choose between a single agent and a multi-agent system

agents, multi-agent, architecture

Prompt

A SaaS company wants to build a system that triages, researches, and proposes fixes for customer-reported bugs, pulling from code, docs, and support tickets, for an engineering org of about 50 people. The team is debating a single well-tooled agent versus an orchestrator with several specialist subagents. Walk through how you'd decide and what you'd build.

Interviewer's answers to clarifying questionsopen after you have asked your own

Volume? ~200 new bug reports/week; the system should propose a draft fix or root-cause summary within an hour for most of them.
Task shape? Some bugs need breadth (search many files/services to find where an error originates); others need a tight, sequential chain (reproduce, then fix, then verify).
Budget? Current single-agent cost per bug is ~$0.80; an early multi-agent prototype costs 4x-15x more per bug.
Review requirement? Every proposed fix goes to a human engineer before merge; nothing auto-merges.
Day-one testing? The team has fewer than 20 hand-labeled example bugs to evaluate against.

Follow-upsafter your high-level design

How would you detect two subagents silently duplicating the same investigation before it wastes budget?
The lead agent, on one bug, spawns 15 subagents for what should have been a 2-subagent job. What's your fallback?
How do you evaluate this system credibly with only 20 labeled examples on day one?
Six months in, the single-agent version your team almost shipped instead would have cost 10x less. Leadership asks whether the multi-agent investment was worth it. What evidence would answer that?

Answer first: out loud, on paper, or with the interview prompt above. Then compare with the rubric.