Learn GenAI
Github
Contents
Map

Quiz · 13 · Agentic Workflows & Multi-Agent Systems

21 questions from 5 pages

These are the Check Yourself questions from each page of the module, collected in course order. Each heading links back to the page the questions test. All module quizzes →

Workflow Patterns

Check yourself
0 / 4 answered
  1. What distinguishes orchestrator-workers from parallelization by sectioning?
  2. When is an evaluator-optimizer loop most likely to improve results?
  3. A router sends 5% of refund requests to the general-FAQ handler. How would you find and fix this?
  4. Best-of-5 with the same model choosing its favourite gives little gain; best-of-5 selected by unit tests gives a large one. Why?

Single-Agent Patterns

Check yourself
0 / 4 answered
  1. Why is model self-reported confidence a poor basis for a confidence gate?
  2. What does using a sub-agent as a tool mainly buy you?
  3. Which verification loop is most likely to fix a bug in generated code?
  4. What should a handoff to a human agent contain?

Multi-Agent Architectures

Check yourself
0 / 5 answered
  1. According to Anthropic's analysis on BrowseComp, what explained most of the performance variance?
  2. Which task is the worst fit for parallel sub-agents?
  3. What is the main difference between handoffs and orchestrator-subagents?
  4. Which MAST category does 'the system stopped before verifying the result' belong to, and what mitigates it?
  5. Why might a multi-agent debate beat a single agent in a paper yet not in your system?

Multi-Agent Engineering

Check yourself
0 / 4 answered
  1. What problem does an artifact store (sub-agents write outputs and return references) solve?
  2. Why keep a structured task ledger in the orchestrator's state?
  3. What is a rainbow deployment for long-running agents?
  4. Your orchestrator spawns 12 sub-agents for 'What is the capital of Australia?'. What do you change?

Patterns Under Measurement

Check yourself
0 / 4 answered
  1. Why do strategies use the docstring examples but grading uses HumanEval's hidden tests?
  2. Why report a paired difference rather than two separate pass rates?
  3. self_refine scored lower than single. Does that prove self-review hurts?
  4. Why did test_feedback cost only 5% more model calls than single?