Microsoft's Azure AI Search has launched a public preview of agentic retrieval, significantly improving answer relevance in conversational AI by up to 40% compared to traditional RAG methods. The system analyzes conversation history and breaks down queries into focused subqueries executed concurrently. This new feature is accessible through a new Knowledge Agents object in the 2025-05-01-preview data plane REST API. By orchestrating a multi-turn retrieval process, Microsoft aims to advance knowledge retrieval systems designed for intelligent agents, providing high-quality grounding data for further applications.
The new agentic retrieval capability enhances relevance in conversational AI by up to 40% compared to traditional RAG, marking a significant advancement in intelligent querying.
This process involves analyzing chat history and original queries to plan focused subqueries that execute in parallel, significantly improving retrieval effectiveness.
Collection
[
|
...
]