Skip to main content
  1. Public practice questions
  2. AB-620 Practice Exams | AI Agent Builder Associate
  3. AB-620 Practice Exam 2

Question 15 — AB-620 Practice Exam 2

A team needs to create a reusable Copilot Studio prompt that accepts different text at runtime, produces a structured result, and can be used after validation in topics or other supported experiences. Which authoring approach should the team use? (Select one answer)

  1. Create the prompt from Tools by selecting Add a tool > New tool > Prompt, define the instructions and an input variable, test representative and edge-case inputs, refine as needed, and save the prompt

  2. Create a generative answers node, place the runtime text in its knowledge sources, test one ideal input, and convert the node into a reusable prompt

  3. Create the prompt directly on a topic canvas, embed each runtime value permanently in the instructions, and save a separate prompt for every input

  4. Create a Condition node, define the classification branches as its instructions, test the branches, and publish the condition as a reusable prompt

Correct answers

  1. Create the prompt from Tools by selecting Add a tool > New tool > Prompt, define the instructions and an input variable, test representative and edge-case inputs, refine as needed, and save the prompt

Explanations

Correct:

Create the prompt from Tools by selecting Add a tool > New tool > Prompt, define the instructions and an input variable, test representative and edge-case inputs, refine as needed, and save the prompt

Prompt builder is opened from the Tools area through Add a tool > New tool > Prompt. Input variables act as placeholders for values supplied at runtime, while the prompt instructions remain fixed. Representative samples and edge cases should be tested and the instructions refined before the prompt is saved for reuse.

Wrong:

Create a generative answers node, place the runtime text in its knowledge sources, test one ideal input, and convert the node into a reusable prompt

Generative answers nodes perform knowledge retrieval rather than authoring reusable task prompts. Testing only an ideal input also provides insufficient validation for variable real-world inputs.

Create the prompt directly on a topic canvas, embed each runtime value permanently in the instructions, and save a separate prompt for every input

Reusable prompts use input variables so runtime values can change without rewriting the instructions. Prompt builder, rather than permanent embedding of each value, provides that reusable design.

Create a Condition node, define the classification branches as its instructions, test the branches, and publish the condition as a reusable prompt

A Condition node performs conversation branching. It is not the authoring environment for AI prompts and does not replace Prompt builder.