Phase 14: AI Agents - Post-Quiz

Test your knowledge after completing the AI Agents phase.

Instructions

  • Answer all 10 questions

  • Try to answer without referring to materials

  • Time limit: 20 minutes

  • Compare your score with the pre-quiz!

Questions

Question 1

Which tool schema property tells the LLM WHEN to use a function?

  • A) name

  • B) description

  • C) parameters

  • D) required

Answer B) description

Explanation: The description field explains what the function does and when it should be used, helping the LLM make intelligent tool selection decisions.

Question 2

In the ReAct pattern, what comes after an Action?

  • A) Final Answer

  • B) Another Action

  • C) Observation

  • D) Thought

Answer C) Observation

Explanation: The ReAct loop is: Thought → Action → Observation → (repeat) → Final Answer. Observations provide results of actions.

Question 3

What is the purpose of the max_iterations parameter in agent loops?

  • A) To make the agent faster

  • B) To prevent infinite loops

  • C) To improve accuracy

  • D) To reduce costs

Answer B) To prevent infinite loops

Explanation: max_iterations caps the number of agent steps, preventing the agent from looping forever if it can’t reach a final answer.

Question 4

Which validation is most critical for a date input parameter?

  • A) Check if it’s a string

  • B) Verify the format (YYYY-MM-DD) and logical validity

  • C) Make sure it’s not empty

  • D) Convert to uppercase

Answer B) Verify the format (YYYY-MM-DD) and logical validity

Explanation: Dates need format validation (correct pattern) AND logic validation (not in past if required, start < end, etc.).

Question 5

What’s the main advantage of using enum in parameter schemas?

  • A) Faster execution

  • B) Limits choices to valid options

  • C) Reduces token usage

  • D) Improves accuracy

Answer B) Limits choices to valid options

Explanation: Enums restrict the LLM to a predefined set of valid values, preventing invalid inputs.

Question 6

In a multi-agent system, what role does a “Planner” agent typically have?

  • A) Executing all tasks

  • B) Breaking down complex tasks into steps

  • C) Reviewing outputs for quality

  • D) Managing API calls

Answer B) Breaking down complex tasks into steps

Explanation: Planner agents analyze the overall task and create a step-by-step plan for other agents to execute.

Question 7

What is “short-term memory” in agent systems?

  • A) The model’s training data

  • B) Recent conversation history

  • C) Cached API responses

  • D) User profile data

Answer B) Recent conversation history

Explanation: Short-term memory stores the current conversation context (recent messages) for immediate reference.

Question 8

Why use exponential backoff in retry logic?

  • A) To fail faster

  • B) To give services time to recover between retries

  • C) To save money

  • D) To improve accuracy

Answer B) To give services time to recover between retries

Explanation: Exponential backoff (1s, 2s, 4s, 8s…) prevents overwhelming struggling services with rapid retry attempts.

Question 9

Which LangChain component manages the agent’s decision-making?

  • A) Tools

  • B) Memory

  • C) AgentExecutor

  • D) Chains

Answer C) AgentExecutor

Explanation: The AgentExecutor runs the agent loop, deciding which tools to call and when to stop.

Question 10

What is “self-correction” in the ReAct pattern?

  • A) Fixing syntax errors

  • B) The agent detecting mistakes and trying different approaches

  • C) Error handling in code

  • D) Validating inputs

Answer B) The agent detecting mistakes and trying different approaches

Explanation: Self-correction occurs when the agent sees an error/unexpected result and reasons about how to fix it, then tries again.

Advanced Questions (Bonus)

Question 11

What’s the primary difference between LangChain and LangGraph?

  • A) LangChain is faster

  • B) LangGraph provides graph-based workflow control

  • C) LangChain is newer

  • D) There’s no difference

Answer B) LangGraph provides graph-based workflow control

Explanation: LangGraph extends LangChain with graph-based state management, enabling more complex agent architectures with cycles and conditionals.

Question 12

In agent memory, what is a “vector database” used for?

  • A) Storing images

  • B) Semantic search over past information

  • C) Faster SQL queries

  • D) Caching API calls

Answer B) Semantic search over past information

Explanation: Vector databases store embeddings, allowing semantic similarity search to retrieve relevant past memories.

Question 13

What problem does “token usage tracking” solve?

  • A) Prevents API rate limits

  • B) Monitors costs and optimizes context length

  • C) Improves accuracy

  • D) Speeds up responses

Answer B) Monitors costs and optimizes context length

Explanation: Tracking tokens helps control costs (tokens = money) and prevents context overflow errors.

Question 14

In multi-agent systems, what is “delegation”?

  • A) Deleting agents

  • B) One agent assigning tasks to other agents

  • C) Error handling

  • D) Parallel execution

Answer B) One agent assigning tasks to other agents

Explanation: Delegation allows a coordinator agent to distribute work to specialized agents based on their capabilities.

Question 15

What’s a key limitation of current AI agents?

  • A) They’re too expensive

  • B) They can hallucinate and make mistakes

  • C) They’re too slow

  • D) They require too much code

Answer B) They can hallucinate and make mistakes

Explanation: LLMs can generate incorrect information confidently. Agents need validation, error handling, and human oversight for critical tasks.

Scoring Guide

Basic Questions (1-10)

  • 0-5 correct: Review the notebooks - some concepts need reinforcement

  • 6-7 correct: Good understanding - practice with challenges

  • 8-9 correct: Strong grasp of concepts - ready for production

  • 10 correct: Excellent! You’ve mastered the fundamentals

Advanced Questions (11-15)

  • 0-2 correct: Focus on advanced notebooks and frameworks

  • 3-4 correct: Solid advanced knowledge - keep practicing

  • 5 correct: Expert level - ready to build complex agents!

Compare Your Growth

Pre-Quiz Score: _____ / 10

Post-Quiz Score: _____ / 15

Improvement: Look at questions you got wrong initially - did you get them right this time?

Next Steps

Based on your score:

If you scored 8+/10 on basics:

  • ✅ Complete the assignment (build your own agent)

  • ✅ Try advanced challenges (multi-agent, memory)

  • ✅ Explore LangChain/LangGraph in depth

  • ✅ Build a production agent for a real use case

If you scored 5-7/10:

  • 📚 Review notebooks 2-4 (function calling, ReAct, frameworks)

  • 🛠️ Complete challenges 1-4

  • 💪 Practice with more examples

  • 🔄 Retake quiz in a few days

If you scored below 5/10:

  • 🔄 Re-read notebooks from the beginning

  • ✏️ Type out all code examples (don’t just read)

  • 🤔 Ask questions in Discord/forums

  • 📝 Take notes on key concepts

  • 🎯 Focus on notebooks 1-3 first

Key Concepts to Master

Make sure you understand:

  • Difference between chatbots and agents

  • How function calling works

  • Tool schema design (name, description, parameters)

  • Input validation strategies

  • ReAct pattern (Thought → Action → Observation)

  • Error handling and self-correction

  • Agent memory (short-term vs long-term)

  • Multi-agent coordination

  • Using LangChain for agents

  • Production considerations (logging, rate limiting, caching)

Feedback

Which topics were:

  • Clearest? _______________

  • Most confusing? _______________

  • Most useful? _______________

  • Want to learn more about? _______________

Congratulations on completing the AI Agents phase! 🎉

You now have the skills to build intelligent, autonomous agents that can solve complex, multi-step problems. Time to build something amazing! 🚀