AI Interview Question
All Questions
DEEP EXPLANATION

Design a multi-agent research system (EXPLAINED)

Scenario BasedAI AgentsHard30 min read

Multi-agent orchestration is a senior/staff-level system design question gaining traction at Google DeepMind and Microsoft. Learn the supervisor pattern, shared state management, and how to avoid the 'too many agents' anti-pattern that sinks most candidate answers.

Design a multi-agent research system
AI Agents · System Design

TL;DR — Quick Answer

Orchestrator agent delegates to specialist agents (search, summarize, critique) with shared memory and structured handoffs.

The Interview Question

Design a system where multiple agents collaborate to research a topic and produce a report.

Deep Explanation

Use a supervisor/orchestrator pattern: Planner agent creates research outline → Search agent gathers sources → Analyst agent synthesizes → Critic agent reviews for gaps → Writer agent produces final report.

Shared state in a blackboard or message bus. Define clear agent roles, output schemas, and termination conditions.

Sign in to unlock full answer

Get deep explanations, PDF export & all AI Agents questions

Multi-AgentOrchestrationResearchGoogleMicrosoft