@@ -23,7 +23,7 @@ zone_pivot_groups: search-csharp-python-rest
 >
 > You're responsible for carefully reviewing and testing applications you build in the context of your specific use cases and making all appropriate decisions and customizations. This includes implementing your own responsible AI mitigations, such as metaprompts, content filters, or other safety systems, and ensuring your applications meet appropriate quality, reliability, security, and trustworthiness standards. For more information, see the [Azure AI Search Transparency Note](/azure/foundry/responsible-ai/search/transparency-note).
 
-A *remote SharePoint knowledge source* (preview) uses the [Copilot Retrieval API](/microsoft-365-copilot/extensibility/api/ai-services/retrieval/overview) (preview) to query textual content directly from SharePoint in Microsoft 365. [Knowledge sources](agentic-knowledge-source-overview.md) are created independently, referenced in a [knowledge base](agentic-retrieval-how-to-create-knowledge-base.md), and used as grounding data when the knowledge base is [queried at runtime](agentic-retrieval-how-to-retrieve.md).
+A *remote SharePoint knowledge source* (preview) uses the [Copilot Retrieval API (preview)](/microsoft-365-copilot/extensibility/api/ai-services/retrieval/overview) to query textual content directly from SharePoint in Microsoft 365. [Knowledge sources](agentic-knowledge-source-overview.md) are created independently, referenced in a [knowledge base](agentic-retrieval-how-to-create-knowledge-base.md), and used as grounding data when the knowledge base is [queried at runtime](agentic-retrieval-how-to-retrieve.md).
 
 To limit sites or constrain search, set a [filter expression](#filter-expression-examples) to scope by URLs, date ranges, file types, and other metadata. The caller's identity must be recognized by both the Azure tenant and the Microsoft 365 tenant because the retrieval engine queries SharePoint on behalf of the user.
 

@@ -34,15 +34,15 @@ Azure AI Search supports the following knowledge sources for agentic retrieval w
 |------|-------------|-------------------|
 | [Search index](agentic-knowledge-source-how-to-search-index.md) | Wraps an existing index. | Indexed |
 | [Azure blob](agentic-knowledge-source-how-to-blob.md) | Generates an indexer pipeline from a blob container. | Indexed |
-| [Azure SQL](agentic-knowledge-source-how-to-azure-sql.md) (preview) | Generates an indexer pipeline from an Azure SQL table or view. | Indexed |
-| [File](agentic-knowledge-source-how-to-file.md) (preview) | Uploads files directly to Azure AI Search. | Indexed |
+| [Azure SQL (preview)](agentic-knowledge-source-how-to-azure-sql.md) | Generates an indexer pipeline from an Azure SQL table or view. | Indexed |
+| [File (preview)](agentic-knowledge-source-how-to-file.md) | Uploads files directly to Azure AI Search. | Indexed |
 | [OneLake](agentic-knowledge-source-how-to-onelake.md) | Generates an indexer pipeline from a lakehouse. | Indexed |
-| [Indexed SharePoint](agentic-knowledge-source-how-to-sharepoint-indexed.md) (preview) | Generates an indexer pipeline from a SharePoint site. | Indexed |
-| [Remote SharePoint](agentic-knowledge-source-how-to-sharepoint-remote.md) (preview) | Retrieves content from SharePoint. | Remote |
-| [Fabric Data Agent](agentic-knowledge-source-how-to-fabric-data-agent.md) (preview) | Retrieves answers and embedded resources from a Microsoft Fabric data agent. | Remote |
-| [Fabric Ontology](agentic-knowledge-source-how-to-fabric-ontology.md) (preview) | Retrieves entity- and relationship-based answers from a Microsoft Fabric ontology. | Remote |
-| [MCP server](agentic-knowledge-source-how-to-mcp-server.md) (preview) | Retrieves live, tool-backed results from an external MCP server. | Remote |
-| [Work IQ](agentic-knowledge-source-how-to-work-iq.md) (preview) | Retrieves organizational intelligence from Work IQ. | Remote |
+| [Indexed SharePoint (preview)](agentic-knowledge-source-how-to-sharepoint-indexed.md) | Generates an indexer pipeline from a SharePoint site. | Indexed |
+| [Remote SharePoint (preview)](agentic-knowledge-source-how-to-sharepoint-remote.md) | Retrieves content from SharePoint. | Remote |
+| [Fabric Data Agent (preview)](agentic-knowledge-source-how-to-fabric-data-agent.md) | Retrieves answers and embedded resources from a Microsoft Fabric data agent. | Remote |
+| [Fabric Ontology (preview)](agentic-knowledge-source-how-to-fabric-ontology.md) | Retrieves entity- and relationship-based answers from a Microsoft Fabric ontology. | Remote |
+| [MCP server (preview)](agentic-knowledge-source-how-to-mcp-server.md) | Retrieves live, tool-backed results from an external MCP server. | Remote |
+| [Work IQ (preview)](agentic-knowledge-source-how-to-work-iq.md) | Retrieves organizational intelligence from Work IQ. | Remote |
 | [Web](agentic-knowledge-source-how-to-web.md) | Retrieves real-time grounding data from Microsoft Bing. | Remote |
 
 ### Indexed knowledge sources

@@ -3,7 +3,7 @@ title: Create a Knowledge Base
 description: Learn how to create a knowledge base for agentic retrieval workloads in Azure AI Search.
 ms.service: azure-ai-search
 ms.topic: how-to
-ms.date: 07/20/2026
+ms.date: 08/06/2026
 ai-usage: ai-assisted
 zone_pivot_groups: search-csharp-python-rest
 ---
@@ -359,7 +359,7 @@ The following JSON is an example response for a knowledge base.
 ```
 
 > [!NOTE]
-> The response schema reflects the API version you used to create the knowledge base. A knowledge base created with the generally available 2026-04-01 API version returns a narrower definition than the 2026-05-01-preview. For more information about which properties each version supports, see the next section.
+> The response schema reflects the API version you used to create the knowledge base. A knowledge base created with the generally available 2026-04-01 API version returns a narrower definition than the 2026-05-01-preview. For more information about which properties each version supports, see [Create a knowledge base](#create-a-knowledge-base).
 
 ## Create a knowledge base
 

@@ -161,7 +161,7 @@ This section supplements the [skill reference](cognitive-search-predefined-skill
 
 1. Add templates for OCR and image analysis from the Azure portal, or copy the definitions from the [skill reference](cognitive-search-predefined-skills.md) documentation. Insert them into the skills array of your skillset definition.
 
-1. If necessary, [include a Microsoft Foundry resource key](cognitive-search-attach-cognitive-services.md) in the skillset. Azure AI Search makes calls to a billable Microsoft Foundry resource for OCR and image analysis for transactions that exceed the free limit (20 per indexer per day). Unless you use a keyless connection (preview), the Microsoft Foundry resource must be in the same region as your search service.
+1. If necessary, [include a Microsoft Foundry resource key](cognitive-search-attach-cognitive-services.md) in the skillset. Azure AI Search makes calls to a billable Microsoft Foundry resource for OCR and image analysis for transactions that exceed the free limit (20 per indexer per day). Unless you use a keyless connection, the Microsoft Foundry resource must be in the same region as your search service.
 
 1. If original images are embedded in PDF or application files like PPTX or DOCX, you need to add a Text Merge skill if you want image output and text output together. Working with embedded images is discussed further on in this article.
 

@@ -9,6 +9,7 @@ ms.custom:
 ms.topic: reference
 ms.date: 04/22/2026
 ms.update-cycle: 365-days
+ai-usage: ai-assisted
 ---
 
 # Document Layout skill
@@ -42,7 +43,7 @@ Supported regions vary by modality and how the skill connects to the Azure Docum
 |----------|-------------|
 | [**Import data** wizard](search-import-data-portal.md) | Create an Azure AI Search service and [Azure AI multi-service account](https://portal.azure.com/#create/Microsoft.CognitiveServicesAllInOne) in one of the following regions: East US, West Europe 2, or North Central US. | 
 | Programmatic, using a [Microsoft Foundry resource key](cognitive-search-attach-cognitive-services.md#bill-through-a-keyless-connection) for billing | Create an Azure AI Search service and Microsoft Foundry resource in the same region. The region must support both [Azure AI Search and Azure Document Intelligence](https://azure.microsoft.com/explore/global-infrastructure/products-by-region/table). |
-| Programmatic, using [Microsoft Entra ID authentication (preview)](cognitive-search-attach-cognitive-services.md#bill-through-a-keyless-connection) for billing | No same-region requirement. Create an Azure AI Search service and Microsoft Foundry resource in any region where [each service is available](https://azure.microsoft.com/explore/global-infrastructure/products-by-region/table). |
+| Programmatic, using [Microsoft Entra ID authentication](cognitive-search-attach-cognitive-services.md#bill-through-a-keyless-connection) for billing | No same-region requirement. Create an Azure AI Search service and Microsoft Foundry resource in any region where [each service is available](https://azure.microsoft.com/explore/global-infrastructure/products-by-region/table). |
 
 ## Supported file formats
 

@@ -7,7 +7,7 @@ ms.custom:
   - dev-focus
 ai-usage: ai-assisted
 ms.topic: how-to
-ms.date: 04/24/2026
+ms.date: 08/06/2026
 ---
 
 # Create a hybrid query in Azure AI Search
@@ -320,9 +320,9 @@ We recommend the [latest preview REST API](/rest/api/searchservice/documents/sea
 
 1. Add a `hybridSearch` query parameter object to set the maximum number of documents recalled through the BM25-ranked results of a hybrid query. It has two properties:
 
-   + `maxTextRecallSize` specifies the number of BM25-ranked results to provide to the Reciprocal Rank Fusion (RRF) ranker used in hybrid queries. The default is 1,000. The maximum is 10,000.
+    + `maxTextRecallSize` specifies the number of BM25-ranked results to provide to the Reciprocal Rank Fusion (RRF) ranker used in hybrid queries. The default is 1,000. The maximum is 10,000.
 
-   + `countAndFacetMode` reports the count and facet scope for a hybrid query. The default, `countAllResults`, uses the full hybrid result set, including all documents that match the text query, even if some of those text matches aren't retrieved for RRF ranking because they fall outside the `maxTextRecallSize` window. Use `countRetrievableResults` to scope count and facets to the documents retrieved for ranking, including `maxTextRecallSize` BM25-ranked documents and the `k` vector matches.
+    + `countAndFacetMode` reports the count and facet scope for a hybrid query. The default, `countAllResults`, uses the full hybrid result set, including all documents that match the text query, even if some of those text matches aren't retrieved for RRF ranking because they fall outside the `maxTextRecallSize` window. Use `countRetrievableResults` to scope count and facets to the documents retrieved for ranking, including `maxTextRecallSize` BM25-ranked documents and the `k` vector matches.
 
 1. Set `maxTextRecallSize`:
 
@@ -516,7 +516,7 @@ api-key: {{admin-api-key}}
 
 ### Example: Semantic hybrid search with filter
 
-Here's the last query in the collection. It's the same semantic hybrid query as the previous example, but with a filter.
+This example adds a `ParkingIncluded` filter to a semantic hybrid query.
 
 ```http
 POST https://{{search-service-name}}.search.windows.net/indexes/{{index-name}}/docs/search?api-version=2026-04-01
@@ -551,7 +551,7 @@ api-key: {{admin-api-key}}
 
 **Key points:**
 
-+ The filter mode can affect the number of results available to the semantic reranker. As a best practice, it's smart to give the semantic ranker the maximum number of documents (50). If prefilters or postfilters are too selective, you might be underserving the semantic ranker by giving it fewer than 50 documents to work with.
++ The filter mode can affect the number of results available to the semantic ranker. As a best practice, give the semantic ranker the maximum number of documents (50). If prefilters or postfilters are too selective, you might underserve the semantic ranker by giving it fewer than 50 documents to work with.
 
 + `preFilter` is applied before query execution. If prefilter reduces the search area to 100 documents, the vector query executes over the `DescriptionVector` field for those 100 documents, returning the k=50 best matches. Those 50 matching documents then pass to RRF for merged results, and then to semantic ranker.
 
@@ -595,7 +595,7 @@ Both `k` and `top` are optional. Unspecified, the default number of results in a
 
 If you're using semantic ranker in 2024-05-01-preview or later, it's a best practice to set `k` and `maxTextRecallSize` to sum to at least 50 total.  You can then restrict the results returned to the user with the `top` parameter. 
 
-If you're using semantic ranker in previous APIs do the following:
+If you're using semantic ranker in an API version earlier than 2024-05-01-preview, follow these steps:
 
 + For keyword-only search (no vectors) set `top` to 50
 + For hybrid search set `k` to 50, to ensure that the semantic ranker gets at least 50 results. 

@@ -1,7 +1,7 @@
 ---
 title: RAG and Generative AI
 description: Learn how Azure AI Search supports RAG patterns with agentic retrieval and classic hybrid search to ground LLM responses in your content. Get started today.
-ms.date: 06/08/2026
+ms.date: 08/04/2026
 ms.service: azure-ai-search
 ms.topic: concept-article
 ms.custom:
@@ -31,7 +31,7 @@ Retrieval-augmented generation (RAG) is a pattern that extends LLM capabilities
 
 Azure AI Search provides two approaches designed specifically for these RAG challenges:
 
-- **[Agentic retrieval](#modern-rag-with-agentic-retrieval) (preview)**: A complete RAG pipeline with LLM-assisted query planning, multi-source access, and structured responses optimized for agent consumption.
+- **[Agentic retrieval](#modern-rag-with-agentic-retrieval)**: A complete RAG pipeline with LLM-assisted query planning, multi-source access, and structured responses optimized for agent consumption.
 
 - **[Classic RAG pattern](#classic-rag-pattern-for-azure-ai-search)**: The proven approach using hybrid search and semantic ranking, ideal for simpler requirements or when generally available (GA) features are required.
 
@@ -147,7 +147,7 @@ Agentic retrieval represents the evolution from traditional single-query RAG pat
 - Built-in semantic ranking for optimal relevance
 - Optional answer synthesis that uses an LLM-formulated answer in the query response
 
-You need new objects for this pipeline: one or more knowledge sources, a knowledge base, and the retrieve action that you call from application code, such as a tool that works with your AI agent.
+You need new objects for this pipeline: one or more knowledge sources (each one points to searchable content, such as a search index or a remote SharePoint site), a knowledge base (the queryable object that unifies those sources), and the retrieve action that you call from application code, such as a tool that works with your AI agent.
 
 For new RAG implementations, start with [agentic retrieval](agentic-retrieval-overview.md). For existing solutions, consider migrating to take advantage of improved accuracy and context understanding.
 

@@ -8,8 +8,7 @@ ms.custom:
   - build-2024
   - ignite-2024
 ms.topic: upgrade-and-migration-article
-ms.date: 06/02/2026
-ai-usage: ai-assisted
+ms.date: 06/11/2026
 ---
 
 # Upgrade to the latest REST API in Azure AI Search
@@ -360,7 +359,7 @@ Use the instructions in this section to migrate vector fields, configuration, an
 
     ```http
     {
-        "search": (this parameter is ignored in vector search),
+        "search": "*", //Required by the API but ignored for ranking in vector-only queries
         "vectors": [
           {
             "value": [
@@ -382,7 +381,7 @@ Use the instructions in this section to migrate vector fields, configuration, an
 
     ```http
     {
-      "search": "(this parameter is ignored in vector search)",
+      "search": "*", //Required by the API but ignored for ranking in vector-only queries
       "vectorQueries": [
         {
           "kind": "vector",

@@ -5,7 +5,7 @@ author: mattwojo
 ms.author: mattwoj
 ms.service: azure-ai-search
 ms.topic: how-to
-ms.date: 06/02/2026
+ms.date: 08/06/2026
 ms.update-cycle: 180-days
 ai-usage: ai-assisted
 ---
@@ -200,7 +200,7 @@ When the search service receives a scale request, it:
 
 Scaling a service can take several minutes to several hours, depending on the size of the service and the scope of the request. Backup duration also varies based on the amount of data and number of partitions and replicas.
 
-The preceding steps aren't entirely consecutive. For example, the system starts provisioning when it can safely do so, which could be while backup is winding down.
+The steps for handling a scale request aren't entirely consecutive. For example, the system starts provisioning when it can safely do so, which could be while backup is winding down.
 
 #### Errors during scaling
 

@@ -121,7 +121,7 @@ In most cases, choose a region near you, unless any of the following apply:
 
 1. Do you have business continuity and disaster recovery (BCDR) requirements? Create two or more search services in different Azure regions, each with two or more replicas so that they can be spread across multiple [availability zones](/azure/reliability/reliability-ai-search#availability-zone-support). For example, if you're operating in North America, you might choose East US and West US, or North Central US and South Central US, for each search service. For more information, see [Multi-region deployments in Azure AI Search](search-multi-region.md).
 
-1. Do you need [AI enrichment](cognitive-search-concept-intro.md), [integrated data chunking and vectorization](vector-search-integrated-vectorization.md), or [multimodal search](multimodal-search-overview.md) powered by Foundry Tools? For billing purposes, you must [attach your Microsoft Foundry resource](cognitive-search-attach-cognitive-services.md) to your search service via a keyless connection (preview) or key-based connection. Key-based connections require both services to be in the same region.
+1. Do you need [AI enrichment](cognitive-search-concept-intro.md), [integrated data chunking and vectorization](vector-search-integrated-vectorization.md), or [multimodal search](multimodal-search-overview.md) powered by Foundry Tools? For billing purposes, you must [attach your Microsoft Foundry resource](cognitive-search-attach-cognitive-services.md) to your search service via a keyless connection or key-based connection. Key-based connections require both services to be in the same region.
 
    - Check [Azure AI Search regions](search-region-support.md#azure-public-regions). If you're using OCR, entity recognition, or other skills backed by Azure AI, the **AI enrichment** column indicates whether Azure AI Search and Microsoft Foundry are in the same region.
 

@@ -1,7 +1,7 @@
 ---
 title: Document-Level Access Control
 description: Learn how Azure AI Search enforces document-level access control with security filters, ACLs, RBAC scopes, SharePoint permissions, and Purview sensitivity labels.
-ms.date: 07/07/2026
+ms.date: 08/05/2026
 ms.reviewer: gimondra
 ms.service: azure-ai-search
 ms.topic: concept-article
@@ -169,7 +169,7 @@ After you synchronize labels, two query paths consume the same indexed label met
 
 - **Knowledge sources and agentic retrieval (MCP)**: Set `ingestionPermissionOptions` to include `sensitivityLabel` on the knowledge source. The retrieve action and MCP `knowledge_base_retrieve` tool return per-reference `sensitivityLabelInfo` and response-level `metadata.responseSensitivityLabelInfo` that clients can use for display banners and policy enforcement. For setup, see [Creating knowledge source](agentic-knowledge-source-overview.md#creating-knowledge-sources) and [Inspect sensitivity label metadata in retrieve responses](agentic-retrieval-how-to-retrieve.md#inspect-sensitivity-label-metadata-in-the-response-preview).
 
-If the knowledge source points to a chunked index, such one populated through integrated vectorization or a custom Text Split skill, the skillset must also [project the sensitivity label to each chunk row](search-indexer-sensitivity-labels.md#6-configure-index-projections-in-your-skillset-if-applicable). Without this projection, chunk-level references aren't filtered.
+If the knowledge source points to a chunked index, such as one populated through integrated vectorization or a custom Text Split skill, the skillset must also [project the sensitivity label to each chunk row](search-indexer-sensitivity-labels.md#6-configure-index-projections-in-your-skillset-if-applicable). Without this projection, chunk-level references aren't filtered.
 
 For more information, see [Use Azure AI Search indexers to ingest Microsoft Purview sensitivity labels](search-indexer-sensitivity-labels.md).
 

@@ -5,7 +5,8 @@ ms.service: azure-ai-search
 ms.custom:
   - ignite-2024
 ms.topic: concept-article
-ms.date: 04/22/2026
+ms.date: 08/05/2026
+ai-usage: ai-assisted
 ---
 
 # Features of Azure AI Search
@@ -23,8 +24,8 @@ The following table summarizes features by category. There's feature parity in a
 
 | Category                             | Features |
 |-------------------|----------|
-| Knowledge bases | [**Knowledge bases (preview)**](agentic-retrieval-how-to-create-knowledge-base.md) orchestrate the agentic retrieval pipeline, connecting to your LLM and managing query parameters. A knowledge base references one or more knowledge sources and defines the retrieval behavior. |
-| Knowledge sources | [**Knowledge sources (preview)**](agentic-knowledge-source-overview.md) specify the content used for agentic retrieval. Create knowledge sources from [search indexes](agentic-knowledge-source-how-to-search-index.md), [Azure Blob Storage](agentic-knowledge-source-how-to-blob.md), [OneLake](agentic-knowledge-source-how-to-onelake.md), [SharePoint (indexed)](agentic-knowledge-source-how-to-sharepoint-indexed.md), [SharePoint (remote)](agentic-knowledge-source-how-to-sharepoint-remote.md), or [web (Bing)](agentic-knowledge-source-how-to-web.md). |
+| Knowledge bases | [**Knowledge bases**](agentic-retrieval-how-to-create-knowledge-base.md) orchestrate the agentic retrieval pipeline, connecting to your LLM and managing query parameters. A knowledge base references one or more knowledge sources and defines the retrieval behavior. |
+| Knowledge sources | [**Knowledge sources**](agentic-knowledge-source-overview.md) specify the content used for agentic retrieval. Create knowledge sources from [search indexes](agentic-knowledge-source-how-to-search-index.md), [Azure Blob Storage](agentic-knowledge-source-how-to-blob.md), [OneLake](agentic-knowledge-source-how-to-onelake.md), [SharePoint (indexed)](agentic-knowledge-source-how-to-sharepoint-indexed.md), [SharePoint (remote)](agentic-knowledge-source-how-to-sharepoint-remote.md), or [web (Bing)](agentic-knowledge-source-how-to-web.md). |
 | Query planning | Query planning uses an LLM to analyze conversation context and break down complex questions into focused subqueries. This includes chat history processing, query decomposition, synonym expansion, and spelling correction. |
 | Parallel query execution | Subqueries run simultaneously across all knowledge sources. Each subquery supports keyword, vector, and hybrid search with automatic semantic reranking. |
 | Retrieval reasoning effort | [**Retrieval reasoning effort (preview)**](agentic-retrieval-how-to-set-retrieval-reasoning-effort.md) controls the level of LLM processing in the pipeline. Set to minimal for speed (no LLM), low for balanced processing, or medium for maximum relevance optimization. |
@@ -34,16 +35,16 @@ The following table summarizes features by category. There's feature parity in a
 
 | Category                             | Features |
 |-------------------|----------|
-| Data sources | Search indexes can accept text from any source, provided it's submitted as a JSON document. </br></br> [**Indexers**](search-indexer-overview.md) are a feature that automates data import from supported data sources to extract searchable content in primary data stores. Indexers handle JSON serialization for you and most support some form of change and deletion detection. You can connect to a [variety of data sources](search-data-sources-gallery.md), including [Microsoft OneLake](search-how-to-index-onelake-files.md), [Azure SQL Database](search-how-to-index-sql-database.md), [Azure Cosmos DB](search-how-to-index-cosmosdb-sql.md), or [Azure Blob storage](search-how-to-index-azure-blob-storage.md).  </br></br>[**Logic Apps connectors (preview)**](search-how-to-index-logic-apps.md) give you access to a broader range of data sources, including data on other cloud platforms. This indexing and enrichment pipeline is created in Azure AI Search but managed in Azure Logic Apps. </br></br>[**Knowledge sources (preview)**](agentic-knowledge-source-overview.md) are a feature of agentic retrieval. They're used during indexing and at query time to support a [knowledge base](agentic-retrieval-how-to-create-knowledge-base.md).|
+| Data sources | Search indexes can accept text from any source, provided it's submitted as a JSON document. </br></br> [**Indexers**](search-indexer-overview.md) are a feature that automates data import from supported data sources to extract searchable content in primary data stores. Indexers handle JSON serialization for you and most support some form of change and deletion detection. You can connect to a [variety of data sources](search-data-sources-gallery.md), including [Microsoft OneLake](search-how-to-index-onelake-files.md), [Azure SQL Database](search-how-to-index-sql-database.md), [Azure Cosmos DB](search-how-to-index-cosmosdb-sql.md), or [Azure Blob Storage](search-how-to-index-azure-blob-storage.md).  </br></br>[**Logic Apps connectors**](search-how-to-index-logic-apps.md) give you access to a broader range of data sources, including data on other cloud platforms. This indexing and enrichment pipeline is created in Azure AI Search but managed in Azure Logic Apps. </br></br>[**Knowledge sources**](agentic-knowledge-source-overview.md) are a feature of agentic retrieval. They're used during indexing and at query time to support a [knowledge base](agentic-retrieval-how-to-create-knowledge-base.md).|
 | Hierarchical and nested data structures | [**Complex types**](search-howto-complex-data-types.md) and collections allow you to model virtually any type of JSON structure within a search index. One-to-many and many-to-many cardinality can be expressed natively through collections, complex types, and collections of complex types.|
 | Linguistic analysis | Analyzers are components of classic search used for text processing during indexing and search operations. By default, you can use the general-purpose Standard Lucene analyzer, or override the default with a language analyzer, a custom analyzer that you configure, or another predefined analyzer that produces tokens in the format you require. </br></br>[**Language analyzers**](index-add-language-analyzers.md) from Lucene or Microsoft are used to intelligently handle language-specific linguistics including verb tenses, gender, irregular plural nouns (for example, 'mouse' vs. 'mice'), word decompounding, word-breaking (for languages with no spaces), and more. </br></br>[**Custom lexical analyzers**](index-add-custom-analyzers.md) are used for complex query forms such as phonetic matching and regular expressions.</br></br> |
 
 ## Chat model and agent integration
 
 | Category&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;  | Features |
 |-------------------|----------|
-| Chat completion models used during indexing | [**GenAI prompt skill**](cognitive-search-skill-genai-prompt.md) is a skill that calls a large language model during indexing and provides a prompt that determines the task. You decide what the task is. It might describing an image, summarizing or manipulating content, or any task the model can perform. Output is added as a new field in a searchable index. |
-| Chat completion models used at query time | [**Agentic retrieval (preview)**](agentic-retrieval-overview.md) uses a large language model for query planning, decomposing and paraphrasing complex queries for better query coverage over your index. Responses from agentic retrieval are designed for agent-to-agent workflows. You can pass search results as single large string, which simplifies agent consumption of your proprietary content. The response also includes citations and query execution information. </br></br>[**Answer synthesis (preview)**](agentic-retrieval-how-to-answer-synthesis.md) uses the LLM to generate citation-backed responses from search results returned by agentic retrieval. </br></br>[**RAG patterns**](retrieval-augmented-generation-overview.md) can be implemented using existing capabilities. The ability to [tune for relevance](search-relevance-overview.md) and construct hybrid queries improve the quality of the content sent to chat bots for answer generation. |
+| Chat completion models used during indexing | [**GenAI prompt skill**](cognitive-search-skill-genai-prompt.md) is a skill that calls a large language model during indexing and provides a prompt that determines the task. You decide what the task is. It might involve describing an image, summarizing or manipulating content, or any task the model can perform. Output is added as a new field in a searchable index. |
+| Chat completion models used at query time | [**Agentic retrieval**](agentic-retrieval-overview.md) uses a large language model for query planning, decomposing and paraphrasing complex queries for better query coverage over your index. Responses from agentic retrieval are designed for agent-to-agent workflows. You can pass search results as single large string, which simplifies agent consumption of your proprietary content. The response also includes citations and query execution information. </br></br>[**Answer synthesis (preview)**](agentic-retrieval-how-to-answer-synthesis.md) uses the LLM to generate citation-backed responses from search results returned by agentic retrieval. </br></br>[**RAG patterns**](retrieval-augmented-generation-overview.md) can be implemented using existing capabilities. The ability to [tune for relevance](search-relevance-overview.md) and construct hybrid queries improve the quality of the content sent to chat bots for answer generation. |
 
 ## Applied AI and AI enriched content
 
@@ -71,7 +72,7 @@ The following table summarizes features by category. There's feature parity in a
 | Category&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;  | Features |
 |-------------------|----------|
 |Free-form text search | [**Full-text search**](search-lucene-query-architecture.md) is a primary use case for most search-based apps. Queries can be formulated using a supported syntax. </br></br>[**Simple query syntax**](query-simple-syntax.md) provides logical operators, phrase search operators, suffix operators, precedence operators. </br></br>[**Full Lucene query syntax**](query-lucene-syntax.md) includes all operations in simple syntax, with extensions for fuzzy search, proximity search, term boosting, and regular expressions.|
-| Relevance | [**Simple scoring**](index-add-scoring-profiles.md) is a key benefit of Azure AI Search. Scoring profiles are used to model relevance as a function of values in the documents themselves. For example, you might want newer products or discounted products to appear higher in the search results. You can also build scoring profiles using tags for personalized scoring based on customer search preferences you've tracked and stored separately. </br></br>[**Semantic ranker**](semantic-search-overview.md) is premium feature that reranks results based on semantic relevance to the query. Depending on your content and scenario, it can significantly improve search relevance with almost minimal configuration or effort. |
+| Relevance | [**Simple scoring**](index-add-scoring-profiles.md) is a key benefit of Azure AI Search. Use scoring profiles to model relevance as a function of values in the documents themselves. For example, you might want newer products or discounted products to appear higher in the search results. You can also build scoring profiles by using tags for personalized scoring based on customer search preferences you track and store separately. </br></br>[**Semantic ranker**](semantic-search-overview.md) is a premium feature that reranks results based on semantic relevance to the query. Depending on your content and scenario, it can significantly improve search relevance with minimal configuration or effort. |
 | Geospatial search | [**Geospatial functions**](search-query-odata-geo-spatial-functions.md) filter over and match on geographic coordinates. You can [match on distance](search-query-simple-examples.md#example-6-geospatial-search) or by inclusion in a polygon shape. |
 | Filters and facets | [**Faceted navigation**](search-faceted-navigation.md) is enabled through a single query parameter. Azure AI Search returns a faceted navigation structure you can use as the code behind a categories list, for self-directed filtering (for example, to filter catalog items by price-range or brand). </br></br> [**Filters**](query-odata-filter-orderby-syntax.md) can be used to incorporate faceted navigation into your application's UI, enhance query formulation, and filter based on user- or developer-specified criteria. Create filters using the OData syntax. |
 | User experience | [**Autocomplete**](search-add-autocomplete-suggestions.md) can be enabled for type-ahead queries in a search bar. </br></br>[**Search suggestions**](/rest/api/searchservice/suggesters) also works off of partial text inputs in a search bar, but the results are actual documents in your index rather than query terms. </br></br>[**Synonyms**](search-synonyms.md) associates equivalent terms that implicitly expand the scope of a query, without the user having to provide the alternate terms. </br></br>[**Hit highlighting**](/rest/api/searchservice/documents/search-post) applies text formatting to a matching keyword in search results. You can choose which fields return highlighted snippets.</br></br>[**Sorting**](/rest/api/searchservice/documents/search-post) is offered for multiple fields via the index schema and then toggled at query-time with a single search parameter.</br></br> [**Paging**](search-pagination-page-layout.md) and throttling your search results is straightforward with the finely tuned control that Azure AI Search offers over your search results.  </br></br>|

@@ -49,7 +49,7 @@ Select the following image to sign in to Azure and open a template. The template
 
 :::image type="content" source="~/reusable-content/ce-skilling/azure/media/template-deployments/deploy-to-azure-button.svg" alt-text="Button to deploy the Resource Manager template to Azure." border="false" link="https://portal.azure.com/#create/Microsoft.Template/uri/https%3A%2F%2Fraw.githubusercontent.com%2Fazure%2Fazure-quickstart-templates%2Fmaster%2Fquickstarts%2Fmicrosoft.search%2Fazure-search-create%2Fazuredeploy.json":::
 
-The Azure portal displays a form that you can use to easily provide parameter values. Some parameters are prefilled with the default values from the template. Provide your subscription, resource group, location, and service name. If you want to use Microsoft Foundry in an [AI enrichment](cognitive-search-concept-intro.md) pipeline, such as for analyzing binary image files for text, choose a location that offers both Azure AI Search and Microsoft Foundry. Unless you use a keyless connection (preview), your Azure AI Search service and Microsoft Foundry resource must be in the same region for AI enrichment workloads. After you complete the form, agree to the terms and conditions, and then select the purchase button to complete your deployment.
+The Azure portal displays a form that you can use to easily provide parameter values. Some parameters are prefilled with the default values from the template. Provide your subscription, resource group, location, and service name. If you want to use Microsoft Foundry in an [AI enrichment](cognitive-search-concept-intro.md) pipeline, such as for analyzing binary image files for text, choose a location that offers both Azure AI Search and Microsoft Foundry. Unless you use a keyless connection, your Azure AI Search service and Microsoft Foundry resource must be in the same region for AI enrichment workloads. After you complete the form, agree to the terms and conditions, and then select the purchase button to complete your deployment.
 
 > [!div class="mx-imgBorder"]
 > ![Azure portal display of template](./media/search-get-started-arm/arm-portalscrnsht.png)

@@ -45,7 +45,7 @@ This article explains how to delete whole documents from a search index on Azure
 
 Manual document deletion is necessary when you use the [push mode approach to indexing](search-what-is-data-import.md#pushing-data-to-an-index), where application code handles data import and drives indexing.
 
-You also need manual document deletion if you use [Logic Apps to load an index (preview)](search-how-to-index-logic-apps.md#limitations).
+You also need manual document deletion if you use [Logic Apps to load an index](search-how-to-index-logic-apps.md#limitations).
 
 You might also need manual document deletion in indexer-driven workloads if search documents become "orphaned" from source documents. An important benefit of indexers is automated content retrieval and synchronization via the change and deletion detection features of the target data source. All of the supported data sources provide some level of detection. But in some cases, synchronized deletion is predicated on a soft-delete strategy where you flag a source document (or record) for deletion, run the indexer to delete the indexed content, and only after the index is updated do you physically delete the source content. If source content is deleted first, you have *orphan documents* in the search index. You must manually delete orphan documents in your index to re-establish parity between source and indexed content.
 

@@ -147,7 +147,7 @@ It also supports the following query actions:
 
 - Deletion detection isn't supported. You must manually [delete orphaned documents](search-how-to-delete-documents.md#delete-a-single-document) from the index.
 
-- Duplicate documents in the search index are a known issue in this preview. Consider deleting objects and starting over if this becomes an issue.
+- Duplicate documents in the search index are a known problem with this feature. If this problem occurs, consider deleting objects and starting over.
 
 - No support for private endpoints in the logic app workflow created by the portal wizard. The workflow is hosted using the [**Consumption** hosting option](/azure/logic-apps/single-tenant-overview-compare) and is subject to its constraints. To use the **Standard** hosting option, use a programmatic approach to creating the workflow.
 

@@ -4,7 +4,8 @@ description: Set up a OneLake indexer to automate indexing of content and metada
 ms.reviewer: gimondra
 ms.service: azure-ai-search
 ms.topic: how-to
-ms.date: 04/23/2026
+ms.date: 08/04/2026
+ai-usage: ai-assisted
 ms.custom:
   - build-2024
   - ignite-2024
@@ -47,7 +48,7 @@ This article uses the REST APIs to illustrate each step.
   + [Use data pipelines](/fabric/data-engineering/tutorial-lakehouse-data-ingestion) from [Microsoft Fabric](https://fabric.microsoft.com/)
   + [Add shortcuts](/fabric/onelake/create-onelake-shortcut) from external data sources like [Amazon S3](/fabric/onelake/create-s3-shortcut) or [Google Cloud Storage](/fabric/onelake/create-gcs-shortcut).  
 
-+ An AI Search service, basic pricing tier or higher, configured for either a [system managed identity](search-how-to-managed-identities.md#create-a-system-managed-identity) or [user-assigned assigned managed identity](search-how-to-managed-identities.md#create-a-user-assigned-managed-identity). The AI Search service must reside within the same tenant as the Microsoft Fabric workspace.
++ An Azure AI Search service, Basic pricing tier or higher, configured for either a [system managed identity](search-how-to-managed-identities.md#create-a-system-managed-identity) or [user-assigned managed identity](search-how-to-managed-identities.md#create-a-user-assigned-managed-identity). The Azure AI Search service must reside within the same tenant as the Microsoft Fabric workspace.
   
 + An Administrator or Contributor role assignment in the Microsoft Fabric workspace where the lakehouse is located. Steps are outlined in the [Grant permissions](#assign-service-permissions) section of this article.
 
@@ -63,13 +64,13 @@ This article uses the REST APIs to illustrate each step.
 
 + This indexer doesn't support OneLake workspace Table location content. 
 
-+ This indexer doesn't support SQL queries, but the query used in the data source configuration is exclusively to add optionally the folder or shortcut to access.
++ This indexer doesn't support SQL queries. The `query` parameter in the data source configuration only specifies an optional folder or shortcut to index.
 
 + There's no support to ingest files from **My Workspace** workspace in OneLake since this is a personal repository per user.
 
 + Indexing files from [Fabric items with sensitivity labels](/fabric/fundamentals/apply-sensitivity-labels), for example, lakehouses, isn't supported. However, when sensitivity labels are applied directly to individual documents, ingestion of protected content and associated labels is supported. In these cases, Azure AI Search can extract and honor sensitivity labels and labeled documents' content through its [integration with Purview](search-indexer-sensitivity-labels.md). 
   
-+ Workspace role-based permissions in Microsoft OneLake may affect indexer access to files. Ensure that the Azure AI Search service principal (managed identity) has sufficient permissions over the files you intend to access in the target [Microsoft Fabric workspace](/fabric/fundamentals/workspaces). 
++ Workspace role-based permissions in Microsoft OneLake might affect indexer access to files. Ensure that the Azure AI Search service principal (managed identity) has sufficient permissions over the files you intend to access in the target [Microsoft Fabric workspace](/fabric/fundamentals/workspaces). 
 
 ## Supported tasks
 
@@ -107,7 +108,7 @@ The following OneLake shortcuts are supported by the OneLake files indexer:
 
 ## Prepare data for indexing
 
-Before you set up indexing, review your source data to determine whether any changes should be made to your data in the lakehouse. An indexer can index content from one container at a time. By default, all files in the container are processed. You have several options for more selective processing:
+Before you set up indexing, review your source data to determine whether you need to make any changes to your data in the lakehouse. An indexer can index content from one container (the lakehouse) at a time. By default, the indexer processes all files in the lakehouse. To process files more selectively, consider the following options:
 
 + Place files in a virtual folder. An indexer [data source definition](#define-the-data-source) includes a "query" parameter that can be either a lakehouse subfolder or shortcut. If this value is specified, only those files in the subfolder or shortcut within the lakehouse are indexed.
 
@@ -155,7 +156,7 @@ The OneLake indexer uses token authentication and role-based access for connecti
 
 The minimum role assignment for your search service identity is Contributor.
 
-1. [Configure a system or user-managed identity](search-how-to-managed-identities.md) for your AI Search service.
+1. [Configure a system or user-managed identity](search-how-to-managed-identities.md) for your Azure AI Search service.
 
    The following screenshot shows a system managed identity for a search service named "onelake-demo".
 
@@ -167,7 +168,7 @@ The minimum role assignment for your search service identity is Contributor.
 
 1. [Grant permission for search service access](/fabric/get-started/give-access-workspaces) to the Fabric workspace. The search service makes the connection on behalf of the indexer.
 
-   If you use a system-assigned managed identity, search for the name of the AI Search service. For a user-assigned managed identity, search for the name of the identity resource.
+   If you use a system-assigned managed identity, search for the name of the Azure AI Search service. For a user-assigned managed identity, search for the name of the identity resource.
 
    The following screenshot shows a Contributor role assignment using a system managed identity.
 

@@ -60,7 +60,7 @@ Use one of the following embedding models for integrated vectorization. Deployme
 
 <sup>2</sup> Azure OpenAI resources (with access to embedding models) that were created in the [Microsoft Foundry portal](https://ai.azure.com/?cid=learnDocs) aren't supported. You must create an Azure OpenAI resource in the Azure portal.
 
-<sup>3</sup> For billing purposes, you must [attach your Microsoft Foundry resource](cognitive-search-attach-cognitive-services.md) to your Azure AI Search skillset. Unless you use a [keyless connection](cognitive-search-attach-cognitive-services.md#bill-through-a-keyless-connection) (preview) to create the skillset, both resources must be in the same region.
+<sup>3</sup> For billing purposes, you must [attach your Microsoft Foundry resource](cognitive-search-attach-cognitive-services.md) to your Azure AI Search skillset. Unless you use a [keyless connection](cognitive-search-attach-cognitive-services.md#bill-through-a-keyless-connection) to create the skillset, both resources must be in the same region.
 
 <sup>4</sup> The Azure Vision multimodal embedding model is available in [select regions](/azure/ai-services/computer-vision/overview-image-analysis#region-availability).
 
@@ -527,7 +527,7 @@ To vectorize your chunked content, the skillset needs an embedding skill that po
 
 <!--### [REST](#tab/embedding-skill-rest)-->
 
-1. After the built-in chunking skill in the `skills` array, call the [Azure OpenAI Embedding skill](cognitive-search-skill-azure-openai-embedding.md) or [Azure Vision multimodal embeddings skill](cognitive-search-skill-vision-vectorize.md) (preview). You can paste one of the following definitions.
+1. After the built-in chunking skill in the `skills` array, call the [Azure OpenAI Embedding skill](cognitive-search-skill-azure-openai-embedding.md) or [Azure Vision multimodal embeddings skill (preview)](cognitive-search-skill-vision-vectorize.md). You can paste one of the following definitions.
 
    ```HTTP
         {
@@ -737,7 +737,7 @@ In this section, you enable vectorization at query time by [defining a vectorize
 
 <!--### [REST](#tab/vectorizer-rest)-->
 
-1. Add the [Azure OpenAI vectorizer](vector-search-vectorizer-azure-open-ai.md) or [Azure Vision vectorizer](vector-search-vectorizer-ai-services-vision.md) (preview) after `vectorSearch.profiles`. You can paste one of the following definitions.
+1. Add the [Azure OpenAI vectorizer](vector-search-vectorizer-azure-open-ai.md) or [Azure Vision vectorizer (preview)](vector-search-vectorizer-ai-services-vision.md) after `vectorSearch.profiles`. You can paste one of the following definitions.
 
    ```HTTP
          "profiles": [ ... ],

@@ -48,7 +48,7 @@ For illustration purposes, this article uses the [sample health plan PDFs](https
 
 - A skillset with these two skills:
 
-  - [Document Layout skill](cognitive-search-skill-document-intelligence-layout.md) that splits documents based on paragraph boundaries. If you use [key-based billing](cognitive-search-attach-cognitive-services.md), this skill requires Microsoft Foundry to be in the same region as Azure AI Search for AI enrichment. Region requirements are relaxed for keyless billing (preview).
+  - [Document Layout skill](cognitive-search-skill-document-intelligence-layout.md) that splits documents based on paragraph boundaries. If you use [key-based billing](cognitive-search-attach-cognitive-services.md), this skill requires Microsoft Foundry to be in the same region as Azure AI Search for AI enrichment. Region requirements are relaxed for keyless billing.
 
   - [Azure OpenAI Embedding skill](cognitive-search-skill-azure-openai-embedding.md) that generates vector embeddings. This skill *doesn't* have region requirements.
 

@@ -9,6 +9,7 @@ ms.custom:
   - ignite-2024
   - build-2025
   - sfi-image-nochange
+ai-usage: ai-assisted
 ---
 
 # Import data wizard in the Azure portal
@@ -36,7 +37,7 @@ Built-in sample data for the hotels-sample index is no longer available. However
 
 ### Data sources
 
-The wizard connects to the following data sources through [built-in indexers](search-indexer-overview.md#supported-data-sources) or [Logic Apps connectors](search-how-to-index-logic-apps.md#supported-connectors) (preview).
+The wizard connects to the following data sources through [built-in indexers](search-indexer-overview.md#supported-data-sources) or [Logic Apps connectors](search-how-to-index-logic-apps.md#supported-connectors).
 
 | Data source | Supported | Connection |
 |--|--|--|

@@ -5,7 +5,7 @@ author: mattwojo
 ms.author: mattwoj
 ms.service: azure-ai-search
 ms.topic: limits-and-quotas
-ms.date: 06/02/2026
+ms.date: 08/04/2026
 ms.update-cycle: 180-days
 ai-usage: ai-assisted
 ms.custom:
@@ -144,9 +144,9 @@ Each index supports up to the following number of documents:
 + 288 billion on L1
 + 576 billion on L2
 
-Each document can be up to approximately 16 megabytes in size. The document size limit actually applies to the size of the indexing API request payload, which is 16 megabytes. That payload can be a single document or a batch of documents. For a batch with a single document, the maximum document size is 16 MB of JSON.
+Each document can be up to approximately 16 MB in size. The document size limit actually applies to the size of the indexing API request payload, which is 16 MB. That payload can be a single document or a batch of documents. For a batch with a single document, the maximum document size is 16 MB of JSON.
 
-The document size limit applies to *push mode* indexing that uploads documents to a search service. If you're using an indexer for *pull mode* indexing, your source files can be any file size, subject to [indexer limits](#indexer-limits). For the blob indexer, file size limits are larger for higher tiers. For example, the S1 limit is 128 megabytes, the S2 limit is 256 megabytes, and so forth.
+The document size limit applies to *push mode* indexing that uploads documents to a search service. If you're using an indexer for *pull mode* indexing, your source files can be any file size, subject to [indexer limits](#indexer-limits). For the blob indexer, file size limits are larger for higher tiers. For example, the S1 limit is 128 MB, and the S2 limit is 256 MB.
 
 When you estimate document size, remember to index only the fields that add value to your search scenarios. Exclude source fields that have no purpose in the queries you intend to run.
 

@@ -1,6 +1,6 @@
 ---
 title: Azure CLI Scripts Using az search Module
-description: Create and configure an Azure AI Search service with the Azure CLI. You can scale a service up or down, manage admin and query api-keys, and query for system information.
+description: Create and configure an Azure AI Search service with the Azure CLI. You can scale a service up or down, manage admin and query API keys, and query for system information.
 author: mattwojo
 ms.author: mattwoj
 ms.service: azure-ai-search
@@ -9,7 +9,7 @@ ms.custom:
   - devx-track-azurecli
   - ignite-2023
 ms.topic: how-to
-ms.date: 06/19/2026
+ms.date: 08/05/2026
 ms.update-cycle: 365-days
 ai-usage: ai-assisted
 ---
@@ -33,7 +33,7 @@ Use the [**az search module**](/cli/azure/search) to perform the following tasks
 > * [Scale up or down with replicas and partitions](#scale-replicas-and-partitions)
 > * [Create a shared private link resource](#create-a-shared-private-link-resource)
 
-Occasionally, questions are asked about tasks *not* on the above list.
+The following information answers common questions about tasks that aren't in the preceding list.
 
 You can't change the name or region of a service programmatically or in the Azure portal. Dedicated resources are allocated when a service is created, so changing the underlying hardware (location or node type) requires a new service.
 
@@ -333,7 +333,7 @@ You can only regenerate one at a time, specified as either the `primary` or `sec
 
 As you might expect, if you regenerate keys without updating client code, requests using the old key will fail. Regenerating all new keys doesn't permanently lock you out of your service, and you can still access the service through the Azure portal. After you regenerate primary and secondary keys, you can update client code to use the new keys and operations will resume accordingly.
 
-Values for the API keys are generated by the service. You can't provide a custom key for Azure AI Search to use. Similarly, there's no user-defined name for admin API-keys. References to the key are fixed strings, either `primary` or `secondary`. 
+The service generates values for the API keys. You can't provide a custom key for Azure AI Search to use. Similarly, there's no user-defined name for admin API keys. References to the key are fixed strings, either `primary` or `secondary`. 
 
 ```azurecli-interactive
 az search admin-key renew \

@@ -31,7 +31,7 @@ Use the [**Az.Search** module](/powershell/module/az.search/) to perform the fol
 > * [Scale up or down with replicas and partitions](#scale-replicas-and-partitions)
 > * [Create a shared private link resource](#create-a-shared-private-link-resource)
 
-Occasionally, questions are asked about tasks *not* on the above list.
+The following information answers common questions about tasks that aren't in the preceding list.
 
 You can't change the name or region of a service programmatically or in the Azure portal. Dedicated resources are allocated when a service is created, so changing the underlying hardware (location or node type) requires a new service.
 

@@ -3,8 +3,9 @@ title: Relevance and Ranking Overview
 description: Learn how relevance and ranking work in Azure AI Search. Explore hybrid search, semantic reranking, agentic retrieval, and scoring profiles to improve results.
 ms.service: azure-ai-search
 ms.topic: concept-article
-ms.date: 12/08/2025
+ms.date: 08/06/2026
 ms.update-cycle: 180-days
+ai-usage: ai-assisted
 ---
 
 # Relevance in Azure AI Search
@@ -21,12 +22,12 @@ The true measure of relevance is *how well* a retrieved set of results meets you
 
 In Azure AI Search, two main strategies have emerged as the best approaches for producing highly relevant results.
 
-+ Hybrid search with semantic reranker
-+ Agentic retrieval (preview) with LLM-assisted query planning and answer formulation
++ Hybrid search with semantic ranker
++ Agentic retrieval with LLM-assisted query planning (preview) and answer synthesis (preview)
 
-[Hybrid search (classic)](hybrid-search-overview.md) delivers on relevance by combining the precision of keyword queries and the semantic similarity of vector queries in a search request targeting a single index. Keyword search operates over a verbatim query. Vector search runs an identical query using a vectorized version of the same string. The queries execute in parallel, looking for precise and semantically similar matches. Results are merged, ranked, and then rescored using a semantic ranker that promotes the most relevant matches. Using keyword and vector search *together* offsets the weaknesses of each approach as a standalone solution. Semantic reranker is an extra component that contributes to a better outcome.
+[Hybrid search (classic)](hybrid-search-overview.md) delivers relevance by combining the precision of keyword queries and the semantic similarity of vector queries in a search request targeting a single index. Keyword search operates over a verbatim query. Vector search runs an identical query using a vectorized version of the same string. The queries execute in parallel, looking for precise and semantically similar matches. Results are merged, ranked, and then rescored using a semantic ranker that promotes the most relevant matches. Using keyword and vector search *together* offsets the weaknesses of each approach as a standalone solution. Semantic ranker is an extra component that contributes to a better outcome.
 
-[Agentic retrieval (preview)](agentic-retrieval-overview.md) delivers on relevance through smart integration with LLMs and a knowledge base that defines an entire search domain. The LLM can analyze and transform queries for more effective retrieval. It can decompose complex questions into targeted subqueries, refine vague requests, or generalize narrow ones for broader scope. In a typical agentic retrieval workload, the LLM answers the question using its reasoning power, context from chat history, and retrieval instructions to identify the very best content and use it to best advantage. This combination of LLM-assisted query planning, multi-source knowledge base search, and LLM reasoning is how agentic retrieval returns highly relevant results.
+[Agentic retrieval](agentic-retrieval-overview.md) delivers relevance through LLM-assisted query planning (preview), answer synthesis (preview), and a knowledge base that defines an entire search domain. The LLM can analyze and transform queries for more effective retrieval. It can decompose complex questions into targeted subqueries, refine vague requests, or generalize narrow ones for broader scope. In a typical agentic retrieval workload, the LLM answers the question using its reasoning power, context from chat history, and retrieval instructions to identify the very best content and use it to best advantage. This combination of LLM-assisted query planning, multi-source knowledge base search, and LLM reasoning is how agentic retrieval returns highly relevant results.
 
 Relevance also depends on having grounding data of sufficient quantity and quality. In agentic retrieval, you can list multiple knowledge sources to expand the scope of what's searchable and provide logic for selecting specific ones.
 
@@ -46,8 +47,8 @@ This section describes the levels of scoring operations. For an illustration of
 |-------|-------------|
 | Level&nbsp;1&nbsp;(L1) | Initial search score (`@search.score`). <br>For text queries matching on tokenized strings, results are always initially ranked using the [BM25 ranking algorithm](index-similarity-and-scoring.md). <br>For vector queries, results are ranked using either [Hierarchical Navigable Small World (HNSW) or exhaustive K-nearest neighbor (KNN)](vector-search-ranking.md). Image search or multimodal searches are based on vector queries and scored using the L1 vector ranking algorithms. |
 | Fused&nbsp;L1 | Scoring from multiple queries using the [Reciprocal Ranking Fusion (RRF) algorithm](hybrid-search-ranking.md). RRF is used for hybrid queries that include text and vector components. RRF is also used when multiple vector queries execute in parallel. A search score from RRF is reflected in `@search.score` [over a different range](#types-of-search-scores).|
-| Level&nbsp;2&nbsp;(L2) | [Semantic ranking score (`@search.reRankerScore`)](semantic-search-overview.md) applies machine reading comprehension to the textual content retrieved by L1 ranking, rescoring the L1 results to better match the semantic intent of the query. L2 reranks L1 results because doing so saves time and money; it would be prohibitive to use semantic ranking as an L1 ranking system. Semantic ranking is a premium feature that bills for usage of the semantic ranking models. It's optional for text queries and vector queries that contain text, but required for [agentic retrieval (preview)](agentic-retrieval-overview.md). Although agentic retrieval sends multiple queries to the query engine, the ranking algorithm for agentic retrieval is the semantic ranker. |
-| Level&nbsp;3&nbsp;(L3) | Applies to [agentic retrieval (preview)](agentic-retrieval-overview.md) and a `medium` retrieval reasoning effort. L3 ranking refers to *iterative search* and it's invoked when the agentic retrieval engine and LLM agree that a second query pass is needed to return a more relevant result. For more information, see [Iterative search for medium retrieval](agentic-retrieval-how-to-set-retrieval-reasoning-effort.md#iterative-search-for-medium-retrieval). |
+| Level&nbsp;2&nbsp;(L2) | [Semantic ranking score (`@search.reRankerScore`)](semantic-search-overview.md) applies machine reading comprehension to the textual content retrieved by L1 ranking, rescoring the L1 results to better match the semantic intent of the query. L2 reranks L1 results because doing so saves time and money; it would be prohibitive to use semantic ranking as an L1 ranking system. Semantic ranking is a premium feature that bills for usage of the semantic ranking models. It's optional for text queries and vector queries that contain text, but required for [agentic retrieval](agentic-retrieval-overview.md). Although agentic retrieval sends multiple queries to the query engine, the ranking algorithm for agentic retrieval is the semantic ranker. |
+| Level&nbsp;3&nbsp;(L3) | Applies to [agentic retrieval](agentic-retrieval-overview.md) and the `medium` retrieval reasoning effort (preview). L3 ranking refers to *iterative search* and it's invoked when the agentic retrieval engine and LLM agree that a second query pass is needed to return a more relevant result. For more information, see [Iterative search for medium retrieval](agentic-retrieval-how-to-set-retrieval-reasoning-effort.md#iterative-search-for-medium-retrieval). |
 
 ## Relevance tuning
 
@@ -101,7 +102,7 @@ The following diagram illustrates how the ranking algorithms work together.
 
 ## Example query inclusive of all ranking algorithms
 
-A query that generates the previous workflow might look like the following example. This hybrid semantic query is scored using RRF (based on L1 scores for text and vectors), and semantic ranking.
+The following hybrid semantic query demonstrates the ranking workflow in the preceding diagram. The query is scored using RRF (based on L1 scores for text and vectors), followed by semantic ranking.
 
 ```http
 POST https://{{search-service-name}}.search.windows.net/indexes/{{index-name}}/docs/search?api-version=2026-05-01-preview
@@ -123,7 +124,7 @@ POST https://{{search-service-name}}.search.windows.net/indexes/{{index-name}}/d
 }
 ```
 
-A response for the previous query includes the original RRF `@search.core` and the `@search.rerankerScore`.
+A response for the hybrid semantic query includes the original RRF `@search.score` and the `@search.rerankerScore`.
 
 ```json
   "value": [

@@ -4,7 +4,7 @@ description: Learn how to configure security features in Azure AI Search to prot
 ms.service: azure-ai-search
 ms.update-cycle: 180-days
 ms.topic: how-to
-ms.date: 03/30/2026
+ms.date: 08/06/2026
 ai-usage: ai-assisted
 ms.custom: horz-security
 ---
@@ -121,7 +121,7 @@ The workflow for authorizing control plane operations is:
 
 ### Authorize data plane operations
 
-Data plane operations target content hosted on a search service, such as index creation, document loading, and queries. Authorization is available through role-based access control, API keys, or both. For configuration steps, see the previous sections on [role-based access control](#recommended-enable-role-based-access-control) and [API key authentication](#configure-api-key-authentication).
+Data plane operations target content hosted on a search service, such as index creation, document loading, and queries. Authorize these operations through role-based access control, API keys, or both. For configuration steps, see [(Recommended) Enable role-based access control](#recommended-enable-role-based-access-control) and [Configure API key authentication](#configure-api-key-authentication).
 
 ### Grant access to individual indexes
 

@@ -3,7 +3,7 @@ title: Configure Customer-Managed Keys for Azure AI Search
 description: Supplement server-side encryption in Azure AI Search using customer managed keys (CMK) or bring your own keys (BYOK) that you create and manage in Azure Key Vault.
 ms.service: azure-ai-search
 ms.topic: how-to
-ms.date: 07/10/2026
+ms.date: 08/05/2026
 ms.update-cycle: 365-days
 ms.custom:
   - references_regions
@@ -18,7 +18,7 @@ ai-usage: ai-assisted
 Enabling customer‑managed keys (CMK) adds additional security on top of the default encryption at rest when using [Microsoft-managed keys](/azure/security/fundamentals/encryption-atrest#azure-encryption-at-rest-components). When you enable CMK, you control the encryption keys used to protect your data, including the ability to:
 
 - Rotate keys on a customer‑defined schedule
-- Disable or revoke keys to block access to encrypted content *(cached keys may persist for up to 60 minutes)*
+- Disable or revoke keys to block access to encrypted content *(cached keys might persist for up to 60 minutes)*
 - Audit key usage through Azure Key Vault logging
 
 You can create, store, and manage keys by using either:
@@ -781,7 +781,7 @@ The response should include the following statement:
 
 ## Rotate or update encryption keys
 
-Use the following instructions to rotate keys or to migrate from Azure Key Vault to the Hardware Security Model (HSM). 
+Use the following instructions to rotate keys or to migrate from Azure Key Vault to the Hardware Security Module (HSM). 
 
 For key rotation, use the [autorotation capabilities of Azure Key Vault](/azure/key-vault/keys/how-to-configure-key-rotation). If you use autorotation, omit the key version in object definitions. The latest key is used, rather than a specific version.
 

@@ -5,7 +5,8 @@ author: mattwojo
 ms.author: mattwoj
 ms.service: azure-ai-search
 ms.topic: concept-article
-ms.date: 07/13/2026
+ms.date: 08/04/2026
+ai-usage: ai-assisted
 ---
 
 # Choose a pricing model and service tier in Azure AI Search
@@ -28,13 +29,13 @@ Selecting the Dedicated pricing model requires estimating your workload needs an
 
 Selecting the Serverless pricing model does not require selecting a pre-provisioned service tier, but uses consumption-based pricing, so [performance optimization](./serverless-cost-optimization.md) will directly affect cost.
 
-## Serverless (preview)
+## Serverless pricing model (preview)
 
 [!INCLUDE [Serverless preview](./includes/previews/preview-serverless.md)]
 
 The Serverless pricing model is a consumption-based offering that automatically scales compute and storage based on your workload. It eliminates the need to provision capacity upfront, allowing you to pay only for the resources you use.
 
-With the Serverless model, you don’t configure replicas, partitions, or search units. Instead, the service manages capacity dynamically in response to query volume, indexing activity, and workload complexity.
+When you use the Serverless model, you don't configure replicas, partitions, or search units. Instead, the service manages capacity dynamically in response to query volume, indexing activity, and workload complexity.
 
 Billing is based on two primary dimensions:
 
@@ -62,7 +63,7 @@ To learn more, see [Service Limits in Azure AI Search](./search-limits-quotas-ca
 
 For additional large-scale Serverless deployment options, contact Microsoft using the [Azure AI Search Serverless Private Preview Sign-up Form](https://aka.ms/FoundryIQ-serverless-contact).
 
-## Dedicated
+## Dedicated pricing model
 
 The Dedicated pricing model is a provisioned-capacity offering that provides predictable performance and cost by allocating fixed infrastructure to your workload. You configure capacity upfront, allowing the service to handle consistent indexing and query demands with guaranteed resources.
 
@@ -71,7 +72,7 @@ With the Dedicated tiers, you explicitly configure replicas, partitions, and sea
 Billing is based on:
 
 - **[Service tier](#tier-descriptions)**: The pre-selected provisioned capacity.
-- **Search units (SUs)**: The billing unit for Dedicated services, calculated as replicas × partitions. You’re billed at a fixed hourly rate based on the number of search units and selected service tier.
+- **Search units (SUs)**: The billing unit for Dedicated services, calculated as replicas × partitions. You pay a fixed hourly rate based on the number of search units and selected service tier.
 
 This model is designed for workloads with steady, predictable demand, where consistent performance, low latency, and controlled scaling are important.
 
@@ -100,9 +101,9 @@ Some tiers are designed for certain types of work:
 
 There is also a free, limited search service tier:
 
-- **Free** creates a [limited search service](search-limits-quotas-capacity.md#subscription-limits) for small projects, such as tutorials and development. Resources are shared across tenants, and scaling is not supported. Some premium features are unavailable, and the service may be deleted after periods of inactivity. You can only have one free search service per Azure subscription.
+- **Free** creates a [limited search service](search-limits-quotas-capacity.md#subscription-limits) for small projects, such as tutorials and development. Resources are shared across tenants, and scaling isn't supported. Some premium features are unavailable, and the service might be deleted after periods of inactivity. You can only have one free search service per Azure subscription.
 
-Billing rates are shown in the [Azure portal](https://portal.azure.com/auth/login/) when you're creating a new AI Search service in the **Select Pricing Tier** page. 
+You see billing rates in the [Azure portal](https://portal.azure.com/auth/login/) when you create a new Azure AI Search service in the **Select Pricing Tier** page. 
 
 :::image type="content" source="media/search-sku-tier/tiers.png" lightbox="media/search-sku-tier/tiers.png" alt-text="Screenshot of the Azure portal Select a pricing tier chart listing the service tiers and their associated SKU." border="true":::
 

@@ -30,7 +30,7 @@ The free account is active for 30 days and includes credits that allow you to cr
 
 ## Choose a region
 
-You can optionally integrate Azure AI Search with Foundry Tools for [AI enrichment](cognitive-search-concept-intro.md), [integrated vectorization](vector-search-integrated-vectorization.md), and [multimodal search](multimodal-search-overview.md). For billing purposes, you must [attach your Microsoft Foundry resource](cognitive-search-attach-cognitive-services.md) to your search service via a keyless connection (preview) or key-based connection. Key-based connections require both services to be in the same region.
+You can optionally integrate Azure AI Search with Foundry Tools for [AI enrichment](cognitive-search-concept-intro.md), [integrated vectorization](vector-search-integrated-vectorization.md), and [multimodal search](multimodal-search-overview.md). For billing purposes, you must [attach your Microsoft Foundry resource](cognitive-search-attach-cognitive-services.md) to your search service via a keyless connection or key-based connection. Key-based connections require both services to be in the same region.
 
 Before you create resources for a key-based connection, confirm regional support:
 

@@ -4,14 +4,15 @@ description: Learn how Azure AI Search helps you build rich search experiences a
 ms.service: azure-ai-search
 ms.update-cycle: 180-days
 ms.topic: overview
-ms.date: 06/02/2026
+ms.date: 08/05/2026
+ai-usage: ai-assisted
 ---
 
 # What is Azure AI Search?
 
 [!INCLUDE [search-fiq-banner](./includes/search-fiq-banner.md)]
 
-Azure AI Search is a fully managed, cloud-hosted service that connects your data to AI. The service unifies access to enterprise and web content so agents and LLMs can use context, chat history, and multi-source signals to produce reliable, grounded answers.
+Azure AI Search is a fully managed, cloud-hosted service that connects your data to AI. The service unifies access to enterprise and web content so agents and large language models (LLMs) can use context, chat history, and multi-source signals to produce reliable, grounded answers.
 
 Azure AI Search is available in two pricing models:
 
@@ -49,7 +50,7 @@ When you create a search service, the following capabilities are included:
 
 + Easily implement search-related features: relevance tuning, faceted navigation, filters (including geo-spatial search), synonym mapping, and autocomplete.
 
-+ Provide enterprise security, access control, and compliance through Microsoft Entra, Azure Private Link, document-level access control, and role-based access.
++ Provide enterprise security, access control, and compliance through Microsoft Entra ID, Azure Private Link, document-level access control, and role-based access.
 
 + Scale and operate in production with Azure reliability, monitoring and diagnostics (logs, metrics, and alerts), and REST API or SDK tooling for automation.
 

@@ -6,7 +6,7 @@ ms.update-cycle: 180-days
 ms.custom:
   - ignite-2023
 ms.topic: concept-article
-ms.date: 04/24/2026
+ms.date: 08/05/2026
 ai-usage: ai-assisted
 ---
 
@@ -30,7 +30,7 @@ Semantic ranker is a collection of query-side capabilities that improve the qual
 
 Secondary ranking and "answers" apply to the query response. Query rewrite is part of the query request.
 
-Here are the capabilities of the semantic reranker.
+Semantic ranker has the following capabilities:
 
 | Capability | Description |
 |---------|-------------|
@@ -59,16 +59,16 @@ Semantic ranking has three steps:
 
 In semantic ranking, the query subsystem passes search results as an input to summarization and ranking models. Because the ranking models have input size constraints and are processing intensive, search results must be sized and structured (summarized) for efficient handling.
 
-1. The semantic ranker starts with a [BM25-ranked result](index-ranking-similarity.md) from a text query or an [RRF-ranked result](hybrid-search-ranking.md) from a vector or hybrid query. The reranking exercise uses only text. Even if results include more than 50 results, only the top 50 results progress to semantic ranking. Typically, semantic ranking uses informational and descriptive fields.
+1. The semantic ranker starts with a [BM25-ranked result](index-similarity-and-scoring.md) from a text query or an [RRF-ranked result](hybrid-search-ranking.md) from a vector or hybrid query. The reranking exercise uses only text. Even if results include more than 50 results, only the top 50 results progress to semantic ranking. Typically, semantic ranking uses informational and descriptive fields.
 
-1. For each document in the search result, the summarization model accepts up to 2,000 tokens, where a token is approximately 10 characters. The model assembles inputs from the "title", "keyword", and "content" fields listed in the [semantic configuration](semantic-how-to-configure.md). 
+1. For each document in the search result, the summarization model accepts up to 2,000 tokens, where a token is approximately 10 characters. The model assembles inputs from the "title", "keywords", and "content" fields listed in the [semantic configuration](semantic-how-to-configure.md). 
 
 1. The system trims excessively long strings to ensure the overall length meets the input requirements of the summarization step. This trimming exercise is why it's important to add fields to your semantic configuration in priority order. If you have very large documents with text-heavy fields, the system ignores anything after the maximum limit.
 
    | Semantic field | Token limit |
    |-----------|-------------|
    | "title"   | 128 tokens |
-   | "keywords | 128 tokens |
+   | "keywords" | 128 tokens |
    | "content" | remaining tokens |
 
 1. The summarization output is a summary string for each document, composed of the most relevant information from each field. The system sends summary strings to the ranker for scoring, and to machine reading comprehension models for captions and answers.

@@ -1,7 +1,7 @@
 ---
 title: Create a Private Endpoint for a Secure Connection
 description: Set up a private endpoint in a virtual network for a secure client connection to an Azure AI Search service.
-ms.date: 06/08/2026
+ms.date: 08/05/2026
 ms.service: azure-ai-search
 ms.topic: how-to
 ms.custom:
@@ -30,7 +30,7 @@ This article walks you through these steps:
 1. [Create an Azure virtual network](#create-the-virtual-network) (or use an existing one)
 1. [Configure a search service with a private endpoint](#create-a-search-service-with-a-private-endpoint)
 1. [Create an Azure virtual machine](#create-a-virtual-machine) in the same virtual network
-1. [Test the connection](#connect-to-the-vm) from the virtual machine
+1. [Test the connection](#test-connections) from the virtual machine
 
 Private endpoints are provided by [Azure Private Link](/azure/private-link/private-link-overview), as a separate billable service. For more information about costs, see [Azure Private Link pricing](https://azure.microsoft.com/pricing/details/private-link/).
 
@@ -199,11 +199,9 @@ Download and then connect to the virtual machine as follows:
 
 ## Test connections
 
-In this section, you verify private network access to the search service and connect privately to the using the Private Endpoint.
+In this section, you use `nslookup` in PowerShell and a REST client from the virtual machine to confirm the search service resolves to a private IP address and is reachable over the private endpoint.
 
-When the search service endpoint is private, some portal features are disabled. You can view and manage service level settings, but portal access to index data and various other components in the service, such as the index, indexer, and skillset definitions, is restricted for security reasons.
-
-1. In the Remote Desktop of *myVM*, open PowerShell.
+1. In the Remote Desktop of *my-vm*, open PowerShell.
 
 1. Enter `nslookup [search service name].search.windows.net`.
 
@@ -226,21 +224,21 @@ When the search service endpoint is private, some portal features are disabled.
 
 1. Completing the quickstart from the VM is your confirmation that the service is fully operational.
 
-1. Close the remote desktop connection to *myVM*.
+1. Close the remote desktop connection to *my-vm*.
 
 1. To verify that your service isn't accessible on a public endpoint, open a REST client on your local workstation and attempt the first several tasks in the quickstart. If you receive an error that the remote server doesn't exist, you successfully configured a private endpoint for your search service.
 
 <a id="portal-access-private-search-service"></a>
 
 ## Use the Azure portal to access a private search service
 
-When the search service endpoint is private, some portal features are disabled. You can view and manage service level information, but index, indexer, and skillset information are hidden for security reasons.
+When the search service endpoint is private, some portal features are disabled. You can still view and manage service-level settings, but portal access to index data and other components of the service (such as the index, indexer, and skillset definitions) is restricted for security reasons.
 
 To work around this restriction, connect to Azure portal from a browser on a virtual machine inside the virtual network. The Azure portal uses the private endpoint on the connection and gives you visibility into content and operations.
 
 1. Follow the [steps to provision a VM that can access the search service through a private endpoint](#create-virtual-machine-private-endpoint).
 
-1. On a virtual machine in your virtual network, open a browser and sign in to the Azure portal. the Azure portal uses the private endpoint attached to the virtual machine to connect to your search service.
+1. On a virtual machine in your virtual network, open a browser and sign in to the Azure portal. The Azure portal uses the private endpoint attached to the virtual machine to connect to your search service.
 
 ## Disable public network access
 

@@ -6,7 +6,7 @@ ms.update-cycle: 180-days
 ms.custom:
   - build-2024
 ms.topic: how-to
-ms.date: 02/19/2026
+ms.date: 08/05/2026
 ai-usage: ai-assisted
 ---
 
@@ -138,7 +138,10 @@ To define a vectorizer and vector profile in an existing index:
     GET https://my-search-service.search.windows.net/indexes/my-index?api-version=2026-04-01 HTTP/1.1
     Authorization: Bearer <your-access-token> // For API keys, replace this line with api-key: <your-admin-api-key>
     ```
-    
+
+    > [!WARNING]
+    > A retrieved Custom Web API vectorizer contains `<redacted>` for every `httpHeaders` value. When you update the same vectorizer without changing its `name`, `kind`, or `uri`, you can resubmit `<redacted>` for matching existing header names. If you change `uri`, submit actual values for every `httpHeaders` entry. For other update rules, see [Custom Web API vectorizer](vector-search-vectorizer-custom-web-api.md#update-header-values-after-get).
+
 1. Use [Indexes - Create Or Update](/rest/api/searchservice/indexes/create-or-update) (REST API) to update the index definition. Paste the full index definition in the request body.
 
    ```http
@@ -162,14 +165,15 @@ To define a vectorizer and vector profile in an existing index:
               "resourceUri": "https://url.openai.azure.com",
               "deploymentId": "text-embedding-ada-002",
               "modelName": "text-embedding-ada-002",
-              "apiKey": "mytopsecretkey"
+              "apiKey": "<your-azure-openai-api-key>"
             }
           },
           {
             "name": "my_custom_vectorizer",
             "kind": "customWebApi",
-            "customVectorizerParameters": {
-              "uri": "https://my-endpoint",
+            "customWebApiParameters": {
+              "uri": "https://contoso.embeddings.com",
+              "httpMethod": "POST",
               "authResourceId": null,
               "authIdentity": null
             }

@@ -2,7 +2,7 @@
 title: Vector Search Overview
 description: Learn about vector search in Azure AI Search for similarity matching across text, images, and multilingual content using numeric embeddings and vector indexes.
 ms.reviewer: robertlee
-ms.date: 06/08/2026
+ms.date: 08/05/2026
 ms.service: azure-ai-search
 ms.topic: concept-article
 ms.custom:
@@ -90,7 +90,7 @@ Azure AI Search is deeply integrated across the Azure AI platform. The following
 | Foundry Agent Service | In Azure AI Search, you can create an *indexed [knowledge source](agentic-knowledge-source-overview.md)* that points to a search index containing vector fields and a vectorizer. You can then parent the knowledge source to a *[knowledge base](agentic-retrieval-how-to-create-knowledge-base.md)* and [connect the knowledge base to Foundry Agent Service](/azure/ai-foundry/agents/how-to/foundry-iq-connect), providing your agents with vector search results for enhanced knowledge retrieval. |
 | Azure data platforms: Azure Blob Storage, Azure Cosmos DB, Azure SQL, Microsoft OneLake | You can use [indexers](search-indexer-overview.md) to automate data ingestion, and then use [integrated vectorization](vector-search-integrated-vectorization.md) to generate embeddings. Azure AI Search can automatically index vector data from [Azure blob indexers](search-how-to-index-azure-blob-storage.md), [Azure Cosmos DB for NoSQL indexers](search-how-to-index-cosmosdb-sql.md), [Azure Data Lake Storage Gen2](search-how-to-index-azure-data-lake-storage.md), [Azure Table Storage](search-how-to-index-azure-tables.md), and [Microsoft OneLake](search-how-to-index-onelake-files.md). For more information, see [Add vector fields to a search index](vector-search-how-to-create-index.md). |
 
-It's also commonly used in open-source frameworks like [LangChain](https://js.langchain.com/docs/integrations/vectorstores/azure_aisearch).
+Azure AI Search is also commonly used in open-source frameworks like [LangChain](https://js.langchain.com/docs/integrations/vectorstores/azure_aisearch).
 
 ## Related content
 

@@ -6,8 +6,9 @@ ms.service: azure-ai-search
 ms.custom:
   - build-2024
 ms.topic: concept-article
-ms.date: 07/07/2026
+ms.date: 08/05/2026
 ms.update-cycle: 365-days
+ai-usage: ai-assisted
 ---
 
 # Custom Web API vectorizer
@@ -28,7 +29,7 @@ Parameters are case sensitive.
 |--------------------|-------------|
 | `uri` | The URI of the Web API to which the JSON payload is sent. Only the **https** URI scheme is allowed. When you retrieve the index with GET, the service returns the `?code=` query parameter value as `?code=<redacted>` to prevent exposure of function keys. To update the vectorizer without changing the stored URI, set `uri` to `<unchanged>`. |
 | `httpMethod` | The method used to send the payload. Allowed methods are `PUT` or `POST`. |
-| `httpHeaders` | A collection of key-value pairs in which keys represent header names and values represent header values sent to your web API with the payload. The following headers are prohibited in this collection: `Accept`, `Accept-Charset`, `Accept-Encoding`, `Content-Length`, `Content-Type`, `Cookie`, `Host`, `TE`, `Upgrade`, `Via`. When you retrieve the index with GET, the service returns `<redacted>` for all header values to prevent exposure of credentials. To update the vectorizer without changing stored header values, set each value to `<unchanged>`. The service restores the original stored value. |
+| `httpHeaders` | A collection of key-value pairs in which keys are header names and values are sent to your web API. The following headers are prohibited: `Accept`, `Accept-Charset`, `Accept-Encoding`, `Content-Length`, `Content-Type`, `Cookie`, `Host`, `TE`, `Upgrade`, and `Via`. GET returns the sentinel value `<redacted>` for every header value. For update requirements, see [Update header values after GET](#update-header-values-after-get). |
 | `authResourceId` | (Optional) A string that, if set, indicates that this vectorizer uses a managed identity for the connection to the function or app hosting the code. This property takes an application (client) ID or app registration in Microsoft Entra ID in one of these formats: `api://<appId>`, `<appId>/.default`, `api://<appId>/.default`. This value scopes the authentication token retrieved by the query pipeline and sent with the custom web API request to the function or app. Setting this property requires that your search service is [configured for managed identity](search-how-to-managed-identities.md) and your Azure function app is [configured for Microsoft Entra sign-in](/azure/app-service/configure-authentication-provider-aad). |
 | `authIdentity` | (Optional) A user-managed identity used by the search service to connect to the function or app hosting the code. You can use either a [system-managed or user-managed identity](search-how-to-managed-identities.md). To use a system-managed identity, leave `authIdentity` blank. |
 | `timeout` | (Optional) The timeout for the HTTP client making the API call. It must be formatted as an XSD `dayTimeDuration` value (a restricted subset of an [ISO 8601 duration](https://www.w3.org/TR/xmlschema11-2/#dayTimeDuration) value). For example, `PT60S` means 60 seconds. If not set, the default is 30 seconds. The timeout can be between 1 and 230 seconds. |
@@ -48,7 +49,7 @@ The Custom Web API vectorizer supports `text`, `imageUrl`, and `imageBinary` vec
             "uri": "https://contoso.embeddings.com",
             "httpMethod": "POST",
             "httpHeaders": {
-                "api-key": "0000000000000000000000000000000000000"
+                "api-key": "<your-header-value>"
             },
             "timeout": "PT60S",
             "authResourceId": null,
@@ -58,35 +59,70 @@ The Custom Web API vectorizer supports `text`, `imageUrl`, and `imageBinary` vec
 ]
 ```
 
-> [!NOTE]
-> When you retrieve the index by using GET, the service returns `<redacted>` for all `httpHeaders` values in the vectorizer configuration to prevent exposure of credentials. To update the vectorizer without changing stored header values, pass `<unchanged>` for each affected field. The service restores the original stored value.
->
-> The following example shows a GET response for the preceding vectorizer:
->
-> ```json
-> {
->     "name": "my-custom-web-api-vectorizer",
->     "kind": "customWebApi",
->     "customWebApiParameters": {
->         "uri": "https://contoso.embeddings.com",
->         "httpMethod": "POST",
->         "httpHeaders": {
->             "api-key": "<redacted>"
->         },
->         "timeout": "PT60S",
->         "authResourceId": null,
->         "authIdentity": null
->     }
-> }
->
->
-> To update this vectorizer without changing the existing `api-key` value, use `<unchanged>`:
->
-> ```json
-> "httpHeaders": {
->     "api-key": "<unchanged>"
-> }
->
+### Update header values after GET
+
+When you retrieve an index definition, the service returns the sentinel `<redacted>` for every `httpHeaders` value in a Custom Web API vectorizer. For example:
+
+```json
+{
+    "name": "my-custom-web-api-vectorizer",
+    "kind": "customWebApi",
+    "customWebApiParameters": {
+        "uri": "https://contoso.embeddings.com",
+        "httpMethod": "POST",
+        "httpHeaders": {
+            "api-key": "<redacted>"
+        },
+        "timeout": "PT60S",
+        "authResourceId": null,
+        "authIdentity": null
+    }
+}
+```
+
+To reuse the stored `api-key` value, update the same existing vectorizer with the same `name` and `kind`, leave its `uri` unchanged, and resubmit the sentinel for the matching header name:
+
+```json
+{
+    "name": "my-custom-web-api-vectorizer",
+    "kind": "customWebApi",
+    "customWebApiParameters": {
+        "uri": "https://contoso.embeddings.com",
+        "httpMethod": "POST",
+        "httpHeaders": {
+            "api-key": "<redacted>"
+        },
+        "timeout": "PT60S",
+        "authResourceId": null,
+        "authIdentity": null
+    }
+}
+```
+
+With an unchanged `uri`, you can mix `<redacted>` for retained header values with actual replacement values for other existing headers. Provide an actual value for each added or renamed header because the sentinel applies only to an existing header with the same name on the same vectorizer.
+
+If you change the `uri`, provide actual values for every `httpHeaders` entry in the same update. The service doesn't reuse stored values for a different `uri`:
+
+```json
+{
+    "name": "my-custom-web-api-vectorizer",
+    "kind": "customWebApi",
+    "customWebApiParameters": {
+        "uri": "https://new.contoso.embeddings.com",
+        "httpMethod": "POST",
+        "httpHeaders": {
+            "api-key": "<new-header-value>"
+        },
+        "timeout": "PT60S",
+        "authResourceId": null,
+        "authIdentity": null
+    }
+}
+```
+
+If the credentials are unavailable and you must change the `uri`, rotate or regenerate them at the external endpoint. Then submit the new `uri` and header values together.
+
+The `<redacted>` value is a service sentinel, not a credential. It can't create a vectorizer or retrieve or reuse a header value stored for another vectorizer.
 
 ## JSON payload structure
 

@@ -1,7 +1,7 @@
 ---
 title: What's New
 description: Stay up to date with the latest Azure AI Search features, updates, and announcements. Discover new capabilities for search, vector, and AI-powered retrieval.
-ms.date: 06/19/2026
+ms.date: 08/05/2026
 ms.service: azure-ai-search
 ms.topic: overview
 ms.custom:
@@ -34,26 +34,26 @@ Learn about the latest updates to Azure AI Search functionality, docs, and sampl
 | Item | Description |
 |--|--|
 | [Search Service 2026-05-01-preview](/rest/api/searchservice/operation-groups?view=rest-searchservice-2026-05-01-preview&preserve-view=true) | New preview REST API version providing programmatic access to the data plane operations described in this table. |
-| [Serverless pricing model](search-sku-tier.md) (preview) | Serverless is a new consumption-based pricing model that Azure AI Search offers alongside the existing Dedicated (provisioned) tiers. With Serverless, you only pay for the compute and indexed storage that you use, with scale-to-zero when idle and no minimum capacity charge. |
-| [File knowledge source](agentic-knowledge-source-how-to-file.md) (preview) | New indexed knowledge source for uploading files directly to a knowledge base without a separate indexer pipeline. |
-| [Azure SQL knowledge source](agentic-knowledge-source-how-to-azure-sql.md) (preview) | New indexed knowledge source backed by Azure SQL Database. |
-| [Fabric Data Agent knowledge source](agentic-knowledge-source-how-to-fabric-data-agent.md) (preview) | New remote knowledge source backed by a Microsoft Fabric Data Agent, enabling retrieval from Fabric-managed data in agentic workflows. |
-| [Fabric Ontology knowledge source](agentic-knowledge-source-how-to-fabric-ontology.md) (preview) | New remote knowledge source backed by a Microsoft Fabric Ontology, enabling structured knowledge retrieval from Fabric in agentic workflows. |
-| [MCP Server knowledge source](agentic-knowledge-source-how-to-mcp-server.md) (preview) | New remote knowledge source that connects to an external Model Context Protocol (MCP) server, allowing agentic retrieval to draw grounding data from any MCP-compatible tool or service. |
-| [Work IQ knowledge source](agentic-knowledge-source-how-to-work-iq.md) (preview) | New remote knowledge source backed by Work IQ, providing access to Microsoft 365 workplace data for agentic retrieval. |
-| [Retrieve defaults for search index knowledge sources](agentic-knowledge-source-how-to-search-index.md) (preview) | Search index knowledge sources now support persisted retrieve defaults, including a `baseFilter` applied to all retrievals and a runtime `filterAddOn` that composes with the base filter using AND logic. A precedence model governs service defaults, knowledge source defaults, and per-request overrides. |
-| [Image serving for indexed knowledge sources](agentic-knowledge-source-overview.md) (preview) | Retrieved documents from indexed knowledge sources can include image content alongside text in agentic retrieval responses. |
-| [Freshness-aware retrieval for indexed knowledge sources](agentic-retrieval-how-to-configure-freshness.md) (preview) | Configure a freshness policy on indexed knowledge sources to bias retrieval toward recently updated documents. Adjust freshness weighting to balance recency with relevance in agentic workflows. |
-| [Knowledge base GPT-5 and CORS support](agentic-retrieval-how-to-create-knowledge-base.md) (preview) | Knowledge bases now support GPT-5 family models, including `gpt-5.4-mini`, for query planning and response generation. Configure CORS via the new `corsOptions` property to enable direct browser-to-service retrieve calls. |
-| [Optional semantic configuration for agentic retrieval](semantic-how-to-configure.md) (preview) | Starting in the 2026-05-01-preview, a semantic configuration is optional in agentic retrieval flows. Classic semantic search still requires an explicit semantic configuration. |
-| [Retrieve action updates](agentic-retrieval-how-to-retrieve.md) (preview) | New parameters for the retrieve action:<p><ul><li>`knowledgeSourceParams.maxOutputDocuments` and `maxOutputDocuments` cap intermediate and final grounding documents returned.</li><li>`failOnError` marks each knowledge source as required or optional.</li><li>`modelName` appears in activity logs when `includeActivity` is `true`.</li></ul> |
-| [Knowledge base and knowledge source service statistics](vector-search-index-size.md) (preview) | [Get Service Statistics](/rest/api/searchservice/get-service-statistics/get-service-statistics?view=rest-searchservice-2026-05-01-preview&preserve-view=true) now returns `knowledgeBasesCount` and `knowledgeSourcesCount` as additive preview counters. |
-| [Microsoft Purview sensitivity labels in retrieve responses](search-document-level-access-overview.md) (preview) | Knowledge base retrieve responses can include `sensitivityLabelInfo` per reference and a `responseSensitivityLabelInfo` top-level field, surfacing Microsoft Purview sensitivity label metadata alongside each retrieved document. |
-| [APIM support for Azure OpenAI skills and vectorizers](search-how-to-configure-azure-openai-api-management.md) (preview) | The [Azure OpenAI Embedding skill](cognitive-search-skill-azure-openai-embedding.md), [GenAI Prompt skill](cognitive-search-skill-genai-prompt.md), and [Azure OpenAI vectorizer](vector-search-vectorizer-azure-open-ai.md) now accept `azure-api.net` endpoints for routing through Azure API Management. |
-| [Network security perimeter and shared private link for Microsoft Foundry](search-security-network-security-perimeter.md) (preview) | Azure AI Search now supports network security perimeter and shared private link for connections to Microsoft Foundry resources, enabling secure private connectivity for skills, vectorizers, and knowledge bases. |
-| [SharePoint indexer updates](search-how-to-index-sharepoint-online.md) (preview) | The SharePoint indexer adds support for ASPX site pages and SharePoint lists (both with ACL support), per-run incremental ACL sync for items with unique permissions, and site group permissions using the `spg:` prefix with federated credential configuration. The new `metadata_spo_site_asset_item_id` field captures the SharePoint item ID. |
-| [List API paging](search-how-to-page-list-results.md) (preview) | New list operations that support cursor-based paging via `$top`, `$skip`, and a continuation token, allowing you to retrieve large result sets incrementally. |
-| [Azure Content Understanding skill updates](cognitive-search-skill-content-understanding.md) (preview) | The Azure Content Understanding skill now supports semantic chunking, AI-based image descriptions, and knowledge store image projection. |
+| [Serverless pricing model (preview)](search-sku-tier.md) | Serverless is a new consumption-based pricing model that Azure AI Search offers alongside the existing Dedicated (provisioned) tiers. With Serverless, you only pay for the compute and indexed storage that you use, with scale-to-zero when idle and no minimum capacity charge. |
+| [File knowledge source (preview)](agentic-knowledge-source-how-to-file.md) | New indexed knowledge source for uploading files directly to a knowledge base without a separate indexer pipeline. |
+| [Azure SQL knowledge source (preview)](agentic-knowledge-source-how-to-azure-sql.md) | New indexed knowledge source backed by Azure SQL Database. |
+| [Fabric Data Agent knowledge source (preview)](agentic-knowledge-source-how-to-fabric-data-agent.md) | New remote knowledge source backed by a Microsoft Fabric Data Agent, enabling retrieval from Fabric-managed data in agentic workflows. |
+| [Fabric Ontology knowledge source (preview)](agentic-knowledge-source-how-to-fabric-ontology.md) | New remote knowledge source backed by a Microsoft Fabric Ontology, enabling structured knowledge retrieval from Fabric in agentic workflows. |
+| [MCP Server knowledge source (preview)](agentic-knowledge-source-how-to-mcp-server.md) | New remote knowledge source that connects to an external Model Context Protocol (MCP) server, allowing agentic retrieval to draw grounding data from any MCP-compatible tool or service. |
+| [Work IQ knowledge source (preview)](agentic-knowledge-source-how-to-work-iq.md) | New remote knowledge source backed by Work IQ, providing access to Microsoft 365 workplace data for agentic retrieval. |
+| [Retrieve defaults for search index knowledge sources (preview)](agentic-knowledge-source-how-to-search-index.md) | Search index knowledge sources now support persisted retrieve defaults, including a `baseFilter` applied to all retrievals and a runtime `filterAddOn` that composes with the base filter using AND logic. A precedence model governs service defaults, knowledge source defaults, and per-request overrides. |
+| [Image serving for indexed knowledge sources (preview)](agentic-knowledge-source-overview.md) | Retrieved documents from indexed knowledge sources can include image content alongside text in agentic retrieval responses. |
+| [Freshness-aware retrieval for indexed knowledge sources (preview)](agentic-retrieval-how-to-configure-freshness.md) | Configure a freshness policy on indexed knowledge sources to bias retrieval toward recently updated documents. Adjust freshness weighting to balance recency with relevance in agentic workflows. |
+| [Knowledge base GPT-5 and CORS support (preview)](agentic-retrieval-how-to-create-knowledge-base.md) | Knowledge bases now support GPT-5 family models, including `gpt-5.4-mini`, for query planning and response generation. Configure CORS via the new `corsOptions` property to enable direct browser-to-service retrieve calls. |
+| [Optional semantic configuration for agentic retrieval (preview)](semantic-how-to-configure.md) | Starting in the 2026-05-01-preview, a semantic configuration is optional in agentic retrieval flows. Classic semantic search still requires an explicit semantic configuration. |
+| [Retrieve action updates (preview)](agentic-retrieval-how-to-retrieve.md) | New parameters for the retrieve action:<p><ul><li>`knowledgeSourceParams.maxOutputDocuments` and `maxOutputDocuments` cap intermediate and final grounding documents returned.</li><li>`failOnError` marks each knowledge source as required or optional.</li><li>`modelName` appears in activity logs when `includeActivity` is `true`.</li></ul> |
+| [Knowledge base and knowledge source service statistics (preview)](vector-search-index-size.md) | [Get Service Statistics](/rest/api/searchservice/get-service-statistics/get-service-statistics?view=rest-searchservice-2026-05-01-preview&preserve-view=true) now returns `knowledgeBasesCount` and `knowledgeSourcesCount` as additive preview counters. |
+| [Microsoft Purview sensitivity labels in retrieve responses (preview)](search-document-level-access-overview.md) | Knowledge base retrieve responses can include `sensitivityLabelInfo` per reference and a `responseSensitivityLabelInfo` top-level field, surfacing Microsoft Purview sensitivity label metadata alongside each retrieved document. |
+| [APIM support for Azure OpenAI skills and vectorizers (preview)](search-how-to-configure-azure-openai-api-management.md) | The [Azure OpenAI Embedding skill](cognitive-search-skill-azure-openai-embedding.md), [GenAI Prompt skill](cognitive-search-skill-genai-prompt.md), and [Azure OpenAI vectorizer](vector-search-vectorizer-azure-open-ai.md) now accept `azure-api.net` endpoints for routing through Azure API Management. |
+| [Network security perimeter and shared private link for Microsoft Foundry (preview)](search-security-network-security-perimeter.md) | Azure AI Search now supports network security perimeter and shared private link for connections to Microsoft Foundry resources, enabling secure private connectivity for skills, vectorizers, and knowledge bases. |
+| [SharePoint indexer updates (preview)](search-how-to-index-sharepoint-online.md) | The SharePoint indexer adds support for ASPX site pages and SharePoint lists (both with ACL support), per-run incremental ACL sync for items with unique permissions, and site group permissions using the `spg:` prefix with federated credential configuration. The new `metadata_spo_site_asset_item_id` field captures the SharePoint item ID. |
+| [List API paging (preview)](search-how-to-page-list-results.md) | New list operations that support cursor-based paging via `$top`, `$skip`, and a continuation token, allowing you to retrieve large result sets incrementally. |
+| [Azure Content Understanding skill updates (preview)](cognitive-search-skill-content-understanding.md) | The Azure Content Understanding skill now supports semantic chunking, AI-based image descriptions, and knowledge store image projection. |
 
 ## April 2026
 
@@ -80,8 +80,8 @@ Learn about the latest updates to Azure AI Search functionality, docs, and sampl
 | Item | Description |
 |--|--|
 | [Search Management 2026-03-01-preview](/rest/api/searchmanagement/operation-groups?view=rest-searchmanagement-2026-03-01-preview&preserve-view=true) | New preview REST API version providing programmatic access to the control plane operations described in this table. |
-| [Service-level CMK](search-security-manage-encryption-keys.md#enable-service-level-cmk-on-new-objects-by-default-preview) (preview) | For security-conscious customers who want more control over how their data is protected, the new `serviceLevelEncryptionKey` property in the `encryptionWithCmk` configuration helps you enable a customer-managed key (CMK) on all newly created objects by default. It ensures all sensitive data in your search service is protected by a key you control, without having to specify key information each time an object is created. You can later rotate from the default key to one specifically for the object. Note that encryption is set at creation time and can't be added to existing objects. |
-| [`knowledgeRetrieval` property](/rest/api/searchmanagement/services/create-or-update?view=rest-searchmanagement-2026-03-01-preview&preserve-view=true) (preview) | Search Management 2026-03-01-preview adds a `knowledgeRetrieval` property to the search service definition. Currently, this property doesn't affect any data plane versions, including Search Service 2025-11-01-preview. |
+| [Service-level CMK (preview)](search-security-manage-encryption-keys.md#enable-service-level-cmk-on-new-objects-by-default-preview) | For security-conscious customers who want more control over how their data is protected, the new `serviceLevelEncryptionKey` property in the `encryptionWithCmk` configuration helps you enable a customer-managed key (CMK) on all newly created objects by default. It ensures all sensitive data in your search service is protected by a key you control, without having to specify key information each time an object is created. You can later rotate from the default key to one specifically for the object. Note that encryption is set at creation time and can't be added to existing objects. |
+| [`knowledgeRetrieval` property (preview)](/rest/api/searchmanagement/services/create-or-update?view=rest-searchmanagement-2026-03-01-preview&preserve-view=true) | Search Management 2026-03-01-preview adds a `knowledgeRetrieval` property to the search service definition. Currently, this property doesn't affect any data plane versions, including Search Service 2025-11-01-preview. |
 | [**Import data** wizard unification](search-import-data-portal.md) | The **Import data** and **Import data (new)** wizards have been unified into a single **Import data** wizard that supports keyword search, RAG, and multimodal RAG. |
 
 ## February 2026
@@ -134,6 +134,11 @@ Learn about the latest updates to Azure AI Search functionality, docs, and sampl
 | August | Indexers | [Improved indexer runtime tracking information (preview)](search-howto-run-reset-indexers.md#check-indexer-runtime-quota-for-s3-hd-search-services). Applies to Standard 3 High Density (S3 HD) services only. [Get Service Statistics](/rest/api/searchservice/get-service-statistics/get-service-statistics?view=rest-searchservice-2025-08-01-preview&preserve-view=true) response now provides cumulative indexer processing information for the entire service. [Get Status - Indexers](/rest/api/searchservice/indexers/get-status?view=rest-searchservice-2025-08-01-preview&preserve-view=true) provides the same information, but for a specific indexer. |
 | August | Vector search | [Strict postfiltering for vector queries (preview)](vector-search-filters.md). New `strictPostFilter` mode for the `vectorFilterMode` parameter. When specified, filters are applied after the global top-`k` vector results are identified, ensuring that returned documents are a subset of the unfiltered results. |
 | August | Vector search | [Increased maximum dimensions for vector fields](search-limits-quotas-capacity.md#index-limits).  The maximum dimensions per vector field are now `4096`. This update applies to all stable and preview REST API versions that support vectors and doesn't introduce breaking changes. |
+| July | REST API | [Search Management 2025-05-01](/rest/api/searchmanagement/operation-groups?view=rest-searchmanagement-2025-05-01&preserve-view=true). Stable release of the REST APIs for the control plane operations described in this table. For migration guidance, see [Upgrade to the latest REST API in Azure AI Search](search-api-migration.md). |
+| July | General availability | [Service upgrade](search-how-to-upgrade.md).  Now generally available through [Upgrade Service (REST API)](/rest/api/searchmanagement/services/upgrade?view=rest-searchmanagement-2025-05-01&preserve-view=true) and the Azure portal. |
+| July | General availability | [Pricing tier change](search-capacity-planning.md#change-your-pricing-tier).  Now generally available through the `sku` property in [Update Service (REST API)](/rest/api/searchmanagement/services/update?view=rest-searchmanagement-2025-05-01&preserve-view=true) and the Azure portal. |
+| July | General availability | [User-assigned managed identity assignment](search-how-to-managed-identities.md). Now generally available through the `identity` property that associates a user-assigned managed identity with a search service configuration. Only the assignment step, via the [Update Service (REST API)](/rest/api/searchmanagement/services/update?view=rest-searchmanagement-2025-05-01&preserve-view=true) or the Azure portal, is generally available. APIs used for data source or model connections that include a user-assigned managed identity are still in preview. |
+| July | General availability | [Network security perimeter](search-security-network-security-perimeter.md).  Now generally available through the [Azure Virtual Network Manager REST APIs](/rest/api/networkmanager/), which are used to join a search service, and the [Search Management REST APIs](/rest/api/searchmanagement/network-security-perimeter-configurations?view=rest-searchmanagement-2025-05-01&preserve-view=true), which are used to view and synchronize the configuration settings. Portal support for both steps is also generally available. |
 | May | Agentic retrieval | [Agentic retrieval (preview)](agentic-retrieval-overview.md) creates a conversational search experience powered by large language models (LLMs) and your proprietary data. Agentic retrieval breaks down complex user queries into subqueries, runs the subqueries in parallel, and extracts grounding data from documents indexed in Azure AI Search. The output is intended for agents and custom chat solutions. A new [knowledge agent](agentic-retrieval-how-to-create-knowledge-base.md) object is introduced in this preview. Its [response payload](agentic-retrieval-how-to-retrieve.md) is designed for downstream agent and chat model consumption, with full transparency of the query plan and reference data. To get started in the portal, see [Quickstart: Agentic retrieval](search-get-started-agentic-retrieval.md). |
 | May | Vector search | [Multivector support (preview)](vector-search-multi-vector-fields.md). Index multiple child vectors within a single document field. You can now use vector types in nested fields of complex collections, effectively allowing multiple vectors to be associated with a single document.|
 | May | Queries | [Scoring profiles with semantic ranking (preview)​](semantic-how-to-enable-scoring-profiles.md). Semantic ranker adds a new field, `@search.rerankerBoostedScore`, to help you maintain consistent relevance and greater control over final ranking outcomes in your search pipeline. |
@@ -145,22 +150,17 @@ Learn about the latest updates to Azure AI Search functionality, docs, and sampl
 | May | Portal update | Import and vectorize data wizard enhancements. This wizard provides two paths for creating and populating vector indexes: [Retrieval Augmented Generation (RAG)](search-get-started-portal-import-vectors.md) and [Multimodal RAG](search-get-started-portal-image-search.md). Logic apps integration is through the RAG path. |
 | May | Index | [Index "description" support (preview)](agentic-retrieval-how-to-create-index.md#add-a-description). The latest preview API adds a description to an index. Consider a Model Context Protocol (MCP) server that must pick the correct index at run time. The decision can be  based on the description rather than on the index name alone. The description must be human readable and under four thousand characters.|
 | May | REST API | [2025-05-01-preview](/rest/api/searchservice/operation-groups?view=rest-searchservice-2025-05-01-preview&preserve-view=true). New data plane preview REST API version providing programmatic access to the preview features announced in this release. |
-| July | REST API | [Search Management 2025-05-01](/rest/api/searchmanagement/operation-groups?view=rest-searchmanagement-2025-05-01&preserve-view=true). Stable release of the REST APIs for the control plane operations described in this table. For migration guidance, see [Upgrade to the latest REST API in Azure AI Search](search-api-migration.md). |
-| July | General availability | [Service upgrade](search-how-to-upgrade.md).  Now generally available through [Upgrade Service (REST API)](/rest/api/searchmanagement/services/upgrade?view=rest-searchmanagement-2025-05-01&preserve-view=true) and the Azure portal. |
-| July | General availability | [Pricing tier change](search-capacity-planning.md#change-your-pricing-tier).  Now generally available through the `sku` property in [Update Service (REST API)](/rest/api/searchmanagement/services/update?view=rest-searchmanagement-2025-05-01&preserve-view=true) and the Azure portal. |
-| July | General availability | [User-assigned managed identity assignment](search-how-to-managed-identities.md). Now generally available through the `identity` property that associates a user-assigned managed identity with a search service configuration. Only the assignment step, via the [Update Service (REST API)](/rest/api/searchmanagement/services/update?view=rest-searchmanagement-2025-05-01&preserve-view=true) or the Azure portal, is generally available. APIs used for data source or model connections that include a user-assigned managed identity are still in preview. |
-| July | General availability | [Network security perimeter](search-security-network-security-perimeter.md).  Now generally available through the [Azure Virtual Network Manager REST APIs](/rest/api/networkmanager/), which are used to join a search service, and the [Search Management REST APIs](/rest/api/searchmanagement/network-security-perimeter-configurations?view=rest-searchmanagement-2025-05-01&preserve-view=true), which are used to view and synchronize the configuration settings. Portal support for both steps is also generally available. |
+| April | Demo | [RAG Time Journey](https://github.com/microsoft/rag-time). Code and video demonstrations of Retrieval Augmented Generation (RAG) workflows that use Azure AI Search. Segments include fundamentals, patterns and use-cases, [vector indexing at scale](https://github.com/microsoft/rag-time/tree/main/Journey%203%20-%20Optimize%20your%20Vector%20Index%20for%20Scale), and [agentic search](https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/bonus-rag-time-journey-agentic-rag/4404652) where you use an agent to evaluate a result and generate a better answer. |
 | March | Service | [Service upgrade (preview)](search-how-to-upgrade.md). Upgrade your search service to higher storage limits in your region. With a one-time upgrade, you no longer need to recreate your service. Available in [Upgrade Service (2025-02-01-preview)](/rest/api/searchmanagement/services/upgrade?view=rest-searchmanagement-2025-02-01-preview&preserve-view=true) and the Azure portal. |
 | March | Pricing | [Pricing tier change (preview)](search-capacity-planning.md#change-your-pricing-tier). Change the [pricing tier](search-sku-tier.md) of your search service. This provides flexibility to scale storage, increase request throughput, and decrease latency based on your needs. Initially, this preview only supported upgrades between Basic and Standard (S1, S2, and S3) tiers, but starting in July 2025, it supports upgrades *and* downgrades between these tiers. Available in [Update Service (2025-02-01-preview)](/rest/api/searchmanagement/services/update?view=rest-searchmanagement-2025-02-01-preview&preserve-view=true#searchupdateservicewithsku) and the Azure portal. |
 | March | Queries | [Facet hierarchies, aggregations, and facet filters (preview)](search-faceted-navigation-examples.md). New facet query parameters support nested facets. For numeric facetable fields, you can sum the values of each field. You can also specify filters on a facet to add inclusion or exclusion criteria. Available in [Search Documents (2025-03-01-preview)](/rest/api/searchservice/documents/search-post?view=rest-searchservice-2025-03-01-preview&preserve-view=true) and the Azure portal.|
 | March | Vector search | [Rescore vector queries over binary quantization using full precision vectors (preview)](vector-search-how-to-quantization.md#supported-rescoring-techniques). For vector indexes that contain binary quantization, you can rescore query results using a full precision vector query. The query engine uses the dot product of the binary embeddings and the vector query for rescoring, which improves the quality of search results.  Set `enableRescoring` and `discardOriginals` to use this feature, and call the latest preview API version on the request.|
 | March | Queries | [Semantic ranker prerelease models (preview)](semantic-how-to-configure.md#opt-in-for-prerelease-semantic-ranking-models). Opt in to use prerelease semantic ranker models if one happens to be available in your region. Available in [Create or Update Index (2025-03-01-preview)](/rest/api/searchservice/indexes/create-or-update?view=rest-searchservice-2025-03-01-preview&preserve-view=true#semanticconfiguration).|
 | March | REST API | [Search Service REST 2025-03-01-preview](/rest/api/searchservice/operation-groups?view=rest-searchservice-2025-03-01-preview&preserve-view=true). Preview release of REST APIs for data plane operations. Adds support for multi-vector embeddings, hierarchical facets, facet aggregation, and facet filters. |
 | March | REST API | [Search Management 2025-02-01-preview](/rest/api/searchmanagement/operation-groups?view=rest-searchmanagement-2025-02-01-preview&preserve-view=true). Preview release of REST APIs for control plane operations. Adds support for in-place upgrade to higher capacity partitions, in-place upgrade to higher tiers, and Azure Confidential Compute. |
-| April | Demo | [RAG Time Journey](https://github.com/microsoft/rag-time). Code and video demonstrations of Retrieval Augmented Generation (RAG) workflows that use Azure AI Search. Segments include fundamentals, patterns and use-cases, [vector indexing at scale](https://github.com/microsoft/rag-time/tree/main/Journey%203%20-%20Optimize%20your%20Vector%20Index%20for%20Scale), and [agentic search](https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/bonus-rag-time-journey-agentic-rag/4404652) where you use an agent to evaluate a result and generate a better answer. |
 | February | Security | [Customer-managed keys support for Managed HSM](search-security-manage-encryption-keys.md). Use either Azure Key Vault or Azure Key Vault Managed HSM (Hardware Security Module) to store customer-managed keys for extra encryption of sensitive content. |
 
-## Previous year's announcements
+## Pre-2025 announcements
 
 + [2024 announcements](/previous-versions/azure/search/search-whats-new-2024)
 + [2023 announcements](/previous-versions/azure/search/search-whats-new-2023)