✅ Take the Test

Questions

Multiple Choice Test (100 Questions)

Instructions: Choose the best answer for each of the following questions.

1. Why is training an AI agent necessary? A. To give it a personality. B. To provide it with general knowledge. C. To give it context for a company’s processes, data, and goals. D. To make it conscious.

2. Which of the following is NOT a high-level goal of training an AI agent? A. Consistency B. Accuracy C. Creativity D. Capability

3. What are the three core layers of an AI agent's intelligence that are shaped by training? A. System Prompting, Knowledge Base, and Human Feedback. B. Hardware, Software, and Network. C. Autonomy, Persona, and Routing. D. Data, Algorithms, and Models.

4. What is the role of the system prompt in an AI agent? A. To provide the agent with its knowledge base. B. To act as the agent's "rulebook" for its behavior. C. To test the agent's responses. D. To connect the agent to external tools.

5. What is the purpose of the knowledge base (Vector Store) in an AI agent? A. To store the agent's system prompts. B. To store the agent's chat history. C. To store documents and files for the AI to retrieve information from. D. To store the agent's source code.

6. What is the final step in the training process of an AI agent? A. Setting the system prompt. B. Uploading the knowledge base. C. Testing and iterating with human feedback. D. Deploying the agent to production.

7. What is the primary mechanism for controlling and directing an AI agent's autonomy? A. The knowledge base. B. The instructional prompt. C. The user's input. D. The underlying AI model.

8. According to the five-level framework for autonomy, what is the user's role at Level 2? A. Operator B. Collaborator C. Consultant D. Observer

9. At which level of autonomy does the agent operate with full independence, with the user only monitoring its performance? A. Level 2 B. Level 3 C. Level 4 D. Level 5

10. What is the ReAct (Reason and Act) paradigm? A. A method for the agent to think step-by-step. B. A loop of Thought -> Action -> Observation. C. A way to handle errors by raising exceptions. D. A technique for giving the agent a persona.

11. What is a best practice for handling errors in an autonomous agent? A. Raise an exception to stop the agent. B. Have the tool return a descriptive error message as its output. C. Ignore the error and continue. D. Ask the user for help.

12. What are the four pillars of a high-quality instructional prompt? A. Role & Goal, Instructions & Constraints, Tools & Resources, and Output Format. B. Persona, Voice, Tone, and Style. C. Autonomy, Routing, Knowledge, and Tools. D. Data, Model, Prompt, and Output.

13. What is the purpose of the "Role & Goal" pillar in a prompt? A. To define the agent's core identity and purpose. B. To provide step-by-step guidance. C. To outline the available tools. D. To specify the output format.

14. What does the SMART framework stand for in the context of defining an agent's objective? A. Simple, Measurable, Achievable, Relevant, Time-bound. B. Specific, Measurable, Achievable, Relevant, Time-bound. C. Strategic, Measurable, Actionable, Realistic, Time-bound. D. Specific, Meaningful, Actionable, Relevant, Testable.

15. What is a key principle for implementing guardrails in a prompt? A. Use vague and general language. B. Use positive reinforcement only. C. Use explicit negative constraints (e.g., "DO NOT"). D. Avoid setting any boundaries.

16. What is the purpose of using strategic frameworks like SWOT analysis in a prompt? A. To make the prompt longer. B. To guide the agent's analysis and move beyond surface-level data retrieval. C. To confuse the agent. D. To test the agent's knowledge of business terms.

17. What is the difference between persona and voice in an AI agent? A. There is no difference. B. Persona is the "who," and voice is the "how." C. Persona is the "how," and voice is the "who." D. Persona is for the user, and voice is for the agent.

18. What is the main benefit of modular and hierarchical prompting? A. It makes prompts shorter. B. It makes prompts more complex. C. It makes prompts more scalable, maintainable, and reliable. D. It makes prompts harder to read.

19. What is "Red Teaming" in the context of testing instructional prompts? A. A type of A/B test. B. A process of trying to "break" the agent by finding its weaknesses. C. A method for writing prompts. D. A way to measure prompt performance.

20. What are the five pillars of data quality for AI agents? A. Accuracy, Completeness, Consistency, Timeliness, and Relevance. B. Size, Speed, Variety, Veracity, and Value. C. Readability, Usability, Accessibility, Reliability, and Security. D. Structure, Format, Language, Style, and Tone.

21. What is the recommended format for unstructured content like documents and wikis to prepare them for an AI agent's knowledge base? A. JSON B. Markdown C. CSV D. XML

22. What is the purpose of a vector store in an AI agent's architecture? A. To store the agent's source code. B. To store user login credentials. C. To store content as embeddings for semantic search. D. To store the agent's chat history.

23. What is the primary purpose of chunking and segmentation in preparing data for a vector store? A. To make the data harder to read. B. To break down large documents into smaller, more manageable pieces. C. To add more information to the data. D. To encrypt the data.

24. Which of the following is a common chunking strategy? A. Random Chunking B. Alphabetical Chunking C. Recursive Chunking D. Chronological Chunking

25. What is the purpose of adding metadata to data chunks? A. To make the data larger. B. To provide additional context and information for filtering and searching. C. To hide the data. D. To make the data less accurate.

26. What is a key best practice for managing a knowledge base for an AI agent? A. Use inconsistent file naming conventions. B. Use a flat folder structure. C. Use a versioning convention in metadata. D. Never update the knowledge base.

27. What is routing in the context of AI agents? A. The process of the agent moving from one server to another. B. The intelligent process of analyzing a user query to direct it to the most appropriate data source, tool, or agent. C. The process of the user navigating through the agent's user interface. D. The process of the agent routing its own internal traffic.

28. Which of the following is a benefit of using a routing system for an AI agent? A. Decreased accuracy. B. Enhanced scalability. C. Slower response times. D. Decreased flexibility.

29. What is an intent taxonomy? A. A list of all the users of an AI agent. B. A structured, hierarchical classification of all the goals a user might have when interacting with an AI agent. C. A list of all the data sources for an AI agent. D. A list of all the tools an AI agent can use.

30. What is the first step in designing an intent taxonomy? A. Building the hierarchy. B. Defining and documenting intents. C. Brainstorming and data gathering. D. Validation and iteration.

31. What is intent-to-data-source mapping? A. The process of connecting each intent in the taxonomy to the specific data source, tool, or agent that is best equipped to handle it. B. The process of mapping user intents to their geographical location. C. The process of mapping user intents to their demographic information. D. The process of mapping user intents to their purchase history.

32. What is a conflict resolution strategy in the context of AI agent routing? A. A strategy for resolving conflicts between different AI agents. B. A strategy for handling situations where the routing system is unsure about the user's intent. C. A strategy for resolving conflicts between the user and the AI agent. D. A strategy for resolving conflicts between the AI agent and its tools.

33. What is a confidence threshold in the context of AI agent routing? A. A threshold for how confident the user is in the AI agent. B. A pre-defined value used to determine the level of certainty in the intent classification. C. A threshold for how confident the AI agent is in its own abilities. D. A threshold for how confident the developers are in the AI agent.

34. What is hybrid retrieval? A. A retrieval technique that combines keyword search and vector search. B. A retrieval technique that only uses keyword search. C. A retrieval technique that only uses vector search. D. A retrieval technique that uses a hybrid of different AI models.

35. What is intent drift? A. The process of the user's intent drifting away from the topic of conversation. B. The natural process by which the patterns of user queries and their underlying intents change over time. C. The process of the AI agent's intent drifting away from its original purpose. D. The process of the developers' intent for the AI agent drifting over time.

36. What is a key metric to monitor for detecting intent drift? A. The number of users of the AI agent. B. The number of times the AI agent is updated. C. The percentage of unclassified queries. D. The cost of running the AI agent.

37. What is the primary cause of hallucinations in AI agents? A. Poor quality source data. B. The AI agent's creativity. C. The user's imagination. D. The AI agent's sense of humor.

38. What is a strategy for reducing hallucinations in AI agents? A. Using more creative system prompts. B. Setting a higher temperature for the LLM. C. Implementing stricter prompting. D. Using a smaller knowledge base.

39. What is A/B testing in the context of AI agents? A. A method of testing the agent's knowledge of the alphabet. B. A method of comparing two versions of something to determine which one performs better. C. A method of testing the agent's ability to perform two tasks at once. D. A method of testing the agent's personality.

40. What is the first step in the A/B testing workflow? A. Create your variations. B. Split your traffic. C. Formulate a hypothesis. D. Run the test and collect data.

41. What is the core principle of modular design for AI agents? A. To make the agent as complex as possible. B. To break down a complex system into smaller, more manageable parts. C. To make the agent a single, indivisible unit. D. To make the agent harder to maintain.

42. Which of the following is a key module in a modular AI agent architecture? A. The User Interface Module. B. The Orchestration Engine. C. The Marketing Module. D. The Sales Module.

43. What are the three pillars of observability for AI agents? A. Logs, Metrics, and Traces. B. Data, Models, and Prompts. C. Users, Developers, and Managers. D. Hardware, Software, and Network.

44. What is the best practice for the format of logs for AI agents? A. Plain text. B. CSV. C. XML. D. JSON.

45. What is the purpose of an observability dashboard? A. To provide a real-time view of the agent's key metrics. B. To allow users to chat with the agent. C. To allow developers to write code for the agent. D. To allow managers to manage the agent's budget.

46. What is the primary benefit of a well-defined persona for an AI agent? A. It makes the agent more expensive. B. It makes the agent more engaging, trustworthy, and predictable. C. It makes the agent harder to use. D. It makes the agent less consistent.

47. What is the purpose of the raia Academy Simulator? A. To simulate the stock market. B. To automate conversation testing by simulating AI-to-AI dialogues. C. To create 3D simulations for video games. D. To simulate the weather.

48. What is a key principle of effective prompting? A. Be as vague as possible. B. Use as much slang as possible. C. Be specific and provide context. D. Ask for as many things as possible at once.

49. What is a best practice for going live with an AI Agent? A. Launch the agent to the public without any internal testing. B. Keep the launch a secret from everyone. C. Finalize testing before launch and communicate the launch internally or externally. D. Delete all training data after the launch.

50. How is the impact of an AI Agent measured? A. By the number of lines of code. B. By the number of developers on the team. C. By a combination of quantitative and qualitative metrics aligned to the use case. D. By the agent's popularity on social media.

51. What is the primary purpose of the manus-render-diagram utility? A. To convert Markdown files to PDF format. B. To render diagram files to PNG format. C. To transcribe speech to text. D. To create a new React app.

52. Which of the following is NOT a supported action for the shell tool? A. exec B. view C. run D. kill

53. What is the purpose of the range parameter in the file tool's read action? A. To specify the number of times to read the file. B. To specify the line number range to read. C. To specify the character range to read. D. To specify the file size range.

54. Which of the following is a supported search type for the search tool? A. social_media B. video C. news D. shopping

55. What is the purpose of the map tool? A. To create a map of a geographical location. B. To spawn parallel subtasks and aggregate results. C. To map a network drive. D. To create a sitemap for a website.

56. What is the intent parameter used for in the browser tool? A. To specify the user's intention for visiting the URL. B. To specify the user's internet speed. C. To specify the user's browser type. D. To specify the user's location.

57. When should you use the slides tool? A. When you need to create a spreadsheet. B. When you need to create a slide-based presentation. C. When you need to edit a video. D. When you need to write a long document.

58. What is a key takeaway about the future of AI Agent development? A. Most organizations will only ever need one AI Agent. B. AI Agent development is a one-time build. C. A centralized platform is important to manage and scale AI Agent deployments. D. Only developers can build and maintain AI solutions.

59. What is the recommended approach for editing a live AI Agent? A. Edit the live agent directly in production. B. Clone the live agent as a dev version, test changes, and then promote to production. C. Delete the live agent and create a new one from scratch. D. AI agents cannot be edited after they are live.

60. What is the purpose of the gmail tool? A. To manage Google Drive files. B. To handle email management and communication tasks. C. To schedule meetings in Google Calendar. D. To create and edit Google Docs.

61. In the five-level framework of autonomy, what is the agent's role at Level 1? A. A partner that can take on delegated sub-tasks. B. A proactive assistant that seeks guidance. C. Acts as a tool, executing direct commands. D. A fully independent worker.

62. What is the primary purpose of defining a "worldview" for an AI agent in its prompt? A. To give the agent a sense of humor. B. To provide a complete and consistent picture of the agent's world, including its tools and limitations. C. To make the prompt more interesting to read. D. To restrict the agent's access to information.

63. What is the Chain of Thought (CoT) technique? A. A method for the agent to generate creative stories. B. A way to encourage the model to "think out loud" by breaking down a problem into intermediate steps. C. A technique for the agent to access external websites. D. A method for the agent to learn from its mistakes.

64. What is the main purpose of giving an agent an "out" in its prompt? A. To allow the agent to end the conversation at any time. B. To enable the agent to know when not to act or when to escalate to a human. C. To give the agent a way to express its opinions. D. To allow the agent to access the internet.

65. What is the primary function of the "Instructions & Constraints" pillar of a prompt? A. To define the agent's persona. B. To provide specific, step-by-step guidance and establish clear boundaries. C. To list the available tools. D. To specify the output format.

66. What is the main benefit of using negative constraints in a prompt? A. They make the prompt more positive. B. They are more difficult for the agent to understand. C. They create powerful boundaries that the agent is highly motivated to respect. D. They are less effective than positive constraints.

67. What is the purpose of the Core Directives Worksheet? A. To help define the objective and guardrails for a new agent. B. To create a to-do list for the agent. C. To design the agent's user interface. D. To write the agent's code.

68. What is the main idea behind using mental models in prompts? A. To make the prompts more confusing. B. To provide the agent with a framework for thinking and problem-solving. C. To test the agent's memory. D. To make the agent more emotional.

69. What is the ReWOO (Reasoning WithOut Observation) framework? A. A framework where the agent first creates a full plan of action and then executes it. B. A framework where the agent acts first and then thinks about its actions. C. A framework where the agent observes its environment and then reacts. D. A framework where the agent reasons about its actions and then observes the results.

70. What is the primary purpose of defining a persona and voice for an AI agent? A. To make the agent sound more robotic. B. To make the agent more engaging, trustworthy, and predictable. C. To make the agent more difficult to understand. D. To make the agent less consistent.

71. What is the main benefit of hierarchical prompting? A. It makes prompts more disorganized. B. It allows a master prompt to delegate tasks to more specialized sub-prompts. C. It makes prompts harder to read. D. It makes prompts less reusable.

72. What is the main goal of the iterative mindset in prompt engineering? A. To create the perfect prompt on the first try. B. To continuously test, refine, and improve the prompt. C. To write the prompt once and never change it. D. To make the prompt as long as possible.

73. What is the purpose of a Data Quality Assessment Report? A. To list all the data sources for the agent. B. To provide prioritized improvement recommendations for the data. C. To describe the agent's data model. D. To showcase the agent's data visualization capabilities.

74. What is the main goal of the data cleaning and standardization phase? A. To make the data more complex. B. To address identified quality issues systematically. C. To add more data to the knowledge base. D. To make the data less accessible.

75. What is the purpose of the data optimization and enhancement phase? A. To make the data smaller. B. To enhance the data for optimal AI agent performance. C. To make the data harder to understand. D. To make the data less relevant.

76. What is a common data hygiene pitfall? A. Iterative improvement. B. One-time cleaning. C. User feedback integration. D. Regular hygiene cycles.

77. What is the purpose of a YAML header in a file for a vector store? A. To add a title to the file. B. To provide metadata for the file. C. To add a table of contents to the file. D. To add a summary to the file.

78. What is the primary reason for using a routing system in a multi-domain AI agent? A. To make the agent slower. B. To make the agent less accurate. C. To direct queries to the most appropriate data source, tool, or agent. D. To make the agent more expensive.

79. What is the purpose of a "catch-all" or "out-of-scope" intent? A. To classify all user queries. B. To detect new and emerging user intents. C. To confuse the user. D. To make the intent taxonomy more complex.

80. What is the final step in the A/B testing workflow? A. Formulate a hypothesis. B. Create your variations. C. Split your traffic. D. Implement the winner.

81. What is the primary purpose of the google_calendar tool? A. To send emails. B. To manage calendar events. C. To create spreadsheets. D. To browse the web.

82. What is the role of a "user-centered framework" in defining agent autonomy? A. It defines autonomy based on the agent's perspective. B. It characterizes autonomy based on the role the user plays in the interaction. C. It focuses on the technical aspects of autonomy. D. It is not relevant to agent autonomy.

83. Why is it important to treat a prompt as a "configuration file for a person"? A. To make the prompt more personal. B. To be explicit and detailed, reducing ambiguity for the agent. C. To make the prompt more difficult to understand. D. To make the agent more human-like.

84. What is the main advantage of using a multi-faceted mandate in a prompt? A. It makes the prompt shorter. B. It allows the agent to adopt different roles or personas in sequence to handle complex tasks. C. It makes the prompt less structured. D. It is not a recommended practice.

85. What is the purpose of a "Data Hygiene Checklist"? A. To ensure the agent's code is clean. B. To ensure thorough data quality implementation. C. To check the agent's physical hygiene. D. To clean the user's data.

86. What is a key activity in the "Data Inventory" phase of data hygiene? A. To delete all data. B. To catalog all data sources, formats, and volumes. C. To enrich the data. D. To validate the data.

87. What is the primary goal of the "Validation & Testing" phase of data hygiene? A. To verify that cleaned data meets quality standards and performs well. B. To add more data to the knowledge base. C. To make the data more complex. D. To delete the data.

88. What is the recommended way to handle version control for knowledge base documents? A. By changing the file name. B. By using a versioning convention in metadata. C. By creating a new file for each version. D. By not using version control.

89. What is the primary purpose of an intent taxonomy in a routing system? A. To make the routing system more complex. B. To provide a structured, hierarchical classification of all possible user goals. C. To make the routing system less accurate. D. To make the routing system slower.

90. What is the role of the "Execution Layer" in a routing system? A. To parse the user's input. B. To make the routing decision. C. To dispatch the query to the chosen destination. D. To generate the final response.

91. What is a key characteristic of LLM-based routing? A. It is static and rule-based. B. It is highly dynamic and flexible, using an LLM to make routing decisions. C. It is less intelligent than other routing mechanisms. D. It is not a recommended approach.

92. What is the primary purpose of a fallback strategy in a routing system? A. To make the system more likely to fail. B. To provide a backup plan for when the primary routing mechanism fails. C. To make the system more complex. D. To make the system less user-friendly.

93. What is the purpose of the "Reciprocal Rank Fusion (RRF)" algorithm? A. To merge the results of keyword search and vector search in hybrid retrieval. B. To calculate the confidence score of an intent. C. To detect intent drift. D. To measure the hallucination rate.

94. What is a key benefit of hybrid retrieval? A. It is less accurate than using a single retrieval method. B. It is more robust and can handle a wider range of queries. C. It is slower than using a single retrieval method. D. It is more expensive than using a single retrieval method.

95. What is the primary goal of the "Observability Module" in a modular AI agent architecture? A. To make the agent less observable. B. To log, monitor, and trace the agent's behavior. C. To make the agent more difficult to debug. D. To make the agent less reliable.

96. What is the purpose of a "trace" in the context of observability? A. To provide a detailed, time-stamped record of every event. B. To provide a numerical representation of the system's performance. C. To provide a complete record of the entire lifecycle of a single request. D. To provide a summary of the agent's performance.

97. What is the primary benefit of using a modular design for AI agents? A. It makes the agent more difficult to scale. B. It makes the agent less maintainable. C. It allows for specialization, scalability, maintainability, and reusability. D. It makes the agent more monolithic.

98. What is the role of the "Prompt Engine" in a modular AI agent architecture? A. To coordinate the activities of all the other modules. B. To generate the final prompts that are sent to the LLM. C. To store and index documents. D. To execute tool calls and return the results.

99. What is the primary purpose of the "Tool Library" in a modular AI agent architecture? A. To provide a collection of external tools and APIs that the agent can use to perform actions. B. To store the agent's prompts. C. To store the agent's knowledge base. D. To monitor the agent's performance.

100. What is the final step in the adaptation workflow for intent drift? A. Review and analyze the data. B. Update the intent taxonomy. C. Retrain the intent classification model. D. Update the intent-to-data-source mappings.

Answers