Skip to main content
  1. Public practice questions
  2. AB-620 Practice Exams | AI Agent Builder Associate
  3. AB-620 Practice Exam 4

Question 54 — AB-620 Practice Exam 4

A prepared Azure AI Search index uses integrated vectorization. What happens when Copilot Studio sends a natural-language query to that index? (Select one answer)

  1. The query must be embedded by a custom function written inside the Copilot Studio agent.

  2. The same embedding model used for indexed document chunks can vectorize the incoming query at runtime.

  3. Copilot Studio replaces the embedding model selected for the Azure AI Search index.

  4. The indexed document chunks are converted back to keyword-only records before retrieval.

Correct answers

  1. The same embedding model used for indexed document chunks can vectorize the incoming query at runtime.

Explanations

Correct:

The same embedding model used for indexed document chunks can vectorize the incoming query at runtime.

Integrated vectorization keeps the indexing and query pipelines consistent by applying the same embedding model to document chunks and incoming queries. This also reduces the need to implement custom vectorization functions.

Wrong:

The query must be embedded by a custom function written inside the Copilot Studio agent.

Integrated vectorization reduces the need for custom query-vectorization functions because Azure AI Search performs query vectorization using the configured embedding model.

Copilot Studio replaces the embedding model selected for the Azure AI Search index.

The embedding model is part of the Azure AI Search index configuration. Copilot Studio connects to the prepared index rather than substituting its own embedding model.

The indexed document chunks are converted back to keyword-only records before retrieval.

Integrated vectorization preserves vector-based retrieval. It does not convert the prepared vectorized chunks into keyword-only records.