[ ABORT TO HUD ]
SEQ. 1
SEQ. 2
SEQ. 3
Azure AI Search Deep Dive
The Search Engine Behind RAG
Azure AI Search is the recommended search service for Foundry RAG implementations, supporting three search modes:
Search Modes
| Mode | How It Works | Best For |
|---|---|---|
| Keyword (BM25) | Traditional text matching | Exact terms, codes, IDs |
| Vector | Semantic similarity via embeddings | Conceptual queries, natural language |
| Hybrid | Keyword + Vector combined | Production (best overall quality) |
| Semantic Ranking | AI reranker on top of results | Maximum relevance accuracy |
Index Architecture
{
"name": "company-docs-index",
"fields": [
{"name": "id", "type": "Edm.String", "key": true},
{"name": "content", "type": "Edm.String", "searchable": true},
{"name": "contentVector", "type": "Collection(Edm.Single)",
"dimensions": 1536, "vectorSearchProfile": "default"},
{"name": "source", "type": "Edm.String", "filterable": true},
{"name": "title", "type": "Edm.String", "searchable": true}
]
}
🎯 Pro Tip: Always use Hybrid search + Semantic Ranking in production. Hybrid search combines the precision of keyword matching with the conceptual understanding of vector search, and semantic ranking further reorders results for maximum relevance.
FOUNDRY VERIFICATION
QUERY 1 // 1
What search configuration provides the best RAG quality in production?
Keyword only
Vector only
Hybrid search + Semantic Ranking
Full-text search