Multi-Agent Orchestration interview questions

Instead of one agent trying to plan, research, write, and check all at once, you split the work across specialized agents — each with a narrow job, its own instructions, and only the tools it needs — and add a coordinator to route between them. A single agent bloats its prompt and loses focus as the task grows, and one early mistake poisons everything after. Focused roles keep each step short and reliable, and independent steps can run in parallel.

The planner reads the goal and breaks it into an ordered list of steps — it decides what and in what order, but does no hands-on work. The workers (executors) each carry out one step and pass their result forward. The orchestrator (supervisor/router) sits in the middle, hands each step to the specialist best equipped for it, collects results, and decides what happens next or whether the job is done.

Planner-executor: one agent makes an ordered plan, others carry out the steps — good for a known pipeline. Supervisor/router: a manager agent reads each task and delegates to the right specialist, then decides what's next — shines when tasks are varied. Debate/critique: agents review each other's work — one drafts, a critic flags errors, sometimes a third votes — the AI version of peer review, worth the extra calls for high-stakes output.