Quick Note — Choose an appropriate method for retrieval and indexing
AI-103 › Unit 1: Plan and manage an Azure AI solution › Choose the appropriate Foundry services for generative AI and agents › Choose an appropriate method for retrieval and indexing
Quick Note — Choose an appropriate method for retrieval and indexing
A team indexes an engineering catalogue with vector search alone because vectors sound modern. Then a user searches BRG-6204 — an exact part number — and gets a list of conceptually similar bearings, none of them the one asked for. Vectors approximate meaning; an identifier is a token to match. The fix is not a better embedding model, it is hybrid search: run both and fuse the results.
Retrieval target
- Choose an appropriate method for retrieval and indexing
- Closed-book recall
- 5 minutes
- Open Microsoft Learn after a miss
Decision anchors
| Prompt | Compact answer |
|---|---|
| Classic search vs agentic retrieval | Classic search — index-first, one query against one index, one request-response cycle, no LLM-assisted planning. Agentic retrieval — a query against a knowledge base of one or more knowledge sources, with planning, decomposition into subqueries, parallel retrieval, semantic reranking, and merging. Classic has no region restrictions; agentic does. |
| Hybrid search | Combines full-text with vector search to balance precision and recall. Lexical handles identifiers, codes, and exact phrases; vector handles paraphrase. Any stem naming both query styles is asking for hybrid. |
| Do you need vectors at all? | "LLMs and agents don't require vectors. Only use them if you need similarity search or if you have content that can be homogenized into vectors." Vectors add an embedding stage, storage, and pipeline cost — adopt them for a reason, not by default. |
| Push vs pull | Pull (indexer) when the content is in a supported data source. Push is your only option if the source is unsupported or if index and source must be synchronized in real time. Either way, Azure AI Search can only index JSON documents. |
| Integrated vectorization | The service generates the embeddings in-pipeline, so the application does not embed content at index time or queries at search time. |
| Indexed vs remote knowledge sources | Indexed sources use the same engines as classic search. Remote sources bypass indexing and are queried live — current without an ingestion run, and the route for user-permission inheritance (remote SharePoint is designed for it; a security filter is the fallback elsewhere). |
| Enrichment runs at indexing time | Skillsets chunk, vectorize, and transform during indexing. Changing a skillset means re-running ingestion, not patching the index — plan enrichment before the first full ingest. |
Read the answers once, then cover the right-hand column and reconstruct each one from the prompt. A useful answer names the requirement, the capability that satisfies it, and the nearest alternative it rejects. If you can only recognize the answer after seeing it, retrieval is not yet secure.
Ninety-second explanation
Without notes, explain:
- When is push the only permitted ingestion method?
- Why does vector search alone fail on part numbers?
- What does a knowledge base do that a single index query does not?
- Why is adding chunking to an existing pipeline more expensive than it looks?
- Which retrieval choice is also an access-control decision?
Then check yourself against current Microsoft Learn. Record the missing decision rule, not merely the missed product name, in your error log.
When to go deeper
Go deeper when the design turns on permission inheritance or on agentic retrieval's regional support — both are documented per-source and per-region. Start from Introduction to Azure AI Search.
Source and freshness
Written against current Microsoft Learn documentation for the AI-103 skills measured (16 April 2026). Reviewed 2026-08-20. Microsoft Learn controls changing product contracts — availability, limits, preview status, naming, and retirement dates move, and this note is deliberately compact.