Question

You are building an Azure AI Search index to ground a RAG application. Documents are chunked and embedded at ingestion using integrated vectorization backed by an Azure OpenAI embedding deployment. At query time, the application receives natural-language user questions as plain text and must retrieve the most semantically relevant chunks. You want to avoid writing custom code in the application to convert the incoming query text into a vector before searching. What should you configure so the search service embeds the query text automatically?

Accepted Answer

Add a vectorizer to the vector profile in the index definition so the service calls the embedding model to convert query text into a vector at query time

Answer

Enable semantic ranker on the index so the L2 reranker converts the text query into embeddings before scoring

Answer

Configure a scoring profile that boosts the vector field, allowing the service to interpret raw text queries as vectors

Answer

Set the query type to 'simple' and pass the text query to the vector field, letting the analyzer tokenize it into an embedding