Quiz for LangGraph and Agentic AI module

No.	Training Unit	Lecture	Training content	Question	Level	Mark	Answer	Answer Option A	Answer Option B	Answer Option C	Answer Option D
1	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	ReAct Pattern	What does “ReAct” stand for when describing agentic reasoning loops?	Easy	1	A	Reasoning and Acting.	Reading and Acting.	Researching and Actuating.	Repeating and Acknowledging.
2	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	ReAct Pattern	In a standard ReAct loop, what is the “Observe” step specifically responsible for?	Medium	1	C	Thinking about the next step.	Generating a new plan.	Receiving and analyzing the results from a tool execution.	Synthesizing the final answer for the user.
3	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Multi-Expert Agents	Why use a “Multi-Expert” pattern (specialized tools) over a single generic web search tool?	Medium	1	B	It reduces API costs and improves speed.	It provides specialized domain knowledge and structured reasoning.	It simplifies the prompt engineering process.	It removes the need for a reasoning-capable LLM.
4	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Prebuilt Components	What is a major advantage of using the prebuilt ToolNode in LangGraph?	Medium	1	D	It is the only technical way to invoke tools.	It provides unlimited free API calls.	it executes asynchronous code faster than custom nodes.	It automatically handles tool execution, parsing, and error management.
5	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Coordinator	What is the main responsibility of the “Coordinator” node in a multi-expert system?	Medium	1	A	To analyze the query and route it to the correct expert tool or agent.	To execute the underlying Python logic of the tools.	To only summarize the final answer once all tools finish.	To manage the database connection for persistence.
6	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Control Flow	How do you standardly prevent an agent from entering an infinite ReAct loop?	Easy	1	C	By switching to a faster LLM model.	By disabling the tool execution node.	By implementing a max_iterations check in the routing logic.	By only allowing HumanMessage inputs.
7	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Expert Tools	In a Multi-Expert architecture, how is a specialized “Expert” typically implemented as a tool?	Hard	1	B	As a hardcoded Python logic script.	As a tool that invokes a specialized LLM with a domain-specific system prompt.	As a separate physical server deployment.	As a raw SQL query to a knowledge base.
8	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Message Protocol	What specific content in an AIMessage signals that the agent wants to use a tool?	Medium	1	B	The content string contains the word “tool”.	The tool_calls parameter is populated with tool details.	It is followed immediately by a SystemMessage.	The role is set to “assistant_tool”.
9	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Planning Agent	What is the role of a “Planning Agent” in advanced agentic workflows?	Hard	1	D	To execute the final code.	To store the memory of the conversation.	To check for grammar errors.	To break down a complex query and decide on a sequence of actions for the coordinator.
10	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Reflection	What does the “Reflection” pattern involve in agentic AI?	Medium	1	A	The agent reviewing its own output or steps to identify and fix errors.	The agent showing its thoughts to the user.	The database reflecting the latest state.	The coordinator mirroring the user’s input.
11	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Multi-Expert Agents	Why is Multi-Expert better than Web Search for analysis?	Medium	1	B	Lower token cost.	Consistent reasoning and synthesis.	No knowledge cutoff.	Faster execution.
12	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Coordinator	What does the “Think” phase of the Coordinator LLM involve?	Medium	1	A	Analyzing if help is needed and who to call.	Writing Python code.	Google searching.	Saving checkpoints.
13	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Coordinator	Which method does a Coordinator use to bind tools?	Easy	1	D	app.compile()	workflow.add_node()	llm.invoke()	llm.bind_tools([…])
14	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Control Flow	Which variable tracks reasoning loops in the State?	Easy	1	C	messages	user_id	current_iteration	checkpoint_id
15	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Control Flow	Which condition forces a ReAct agent to END?	Medium	1	A	No tool_calls or max_iterations reached.	User stops typing.	API key expires.	After every tool call.
16	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Coordinator	How does a Coordinator execute parallel consultations?	Hard	1	B	Opening multiple prompts.	Populating multiple items in tool_calls.	Re-running the graph.	Parallel execution is not supported.
17	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Planning Agent	What architectural issue does a Planning Agent solve?	Medium	1	C	High latency.	Local hardware limits.	Coordinator overload from too many tasks.	Bypassing checkpointers.
18	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Multi-Expert Agents	How does dynamic expert selection work in advanced systems?	Hard	1	D	Browser history check.	Random assignment.	Exact keyword mapping.	Routing prompt returns best expert name.
19	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Multi-Expert Agents	What is the purpose of structured synthesis?	Easy	1	A	To combine expert insights into a unified response.	To compress JSON state.	To translate output to code.	To hide tool errors.
20	Unit 2: Agentic Patterns: Multi-Expert Research Agent	Lec2	Multi-Expert Agents	What is a major trade-off of the Multi-Expert Pattern?	Medium	1	B	Lower accuracy.	Higher cost and latency.	No LangGraph support.	Requires C++ code.