more edits

Author: HeidiSteen

articles/search/retrieval-augmented-generation-overview.md

@@ -90,15 +90,15 @@ Since you probably know what kind of content you want to search over, consider t
 | Content type | Indexed as | Features |
 |--------------|------------|----------|
 | text | tokens, unaltered text | [Indexers](search-indexer-overview.md) can pull plain text from other Azure resources like Azure Storage and Cosmos DB. You can also [push any JSON content](search-what-is-data-import.md) to an index. To modify text in flight, use [analyzers](search-analyzers.md) and [normalizers](search-normalizers.md) to add lexical processing during indexing. [Synonym maps](search-synonyms.md) are useful if source documents are missing terminology that might be used in a query. |
-| text | vectors <sup>1</sup> | Text can be chunked and vectorized externally and then [indexed as vector fields](vector-search-how-to-create-index.md) in your index. |
+| text | vectors <sup>1</sup> | Text can be chunked and vectorized in an indexer pipeline, or handled externally and then [indexed as vector fields](vector-search-how-to-create-index.md) in your index. |
 | image | tokens, unaltered text <sup>2</sup> | [Skills](cognitive-search-working-with-skillsets.md) for OCR and Image Analysis can process images for text recognition or image characteristics. Image information is converted to searchable text and added to the index. Skills have an indexer requirement. |
-| image | vectors <sup>1</sup> | Images can be vectorized externally for a mathematical representation of image content and then [indexed as vector fields](vector-search-how-to-create-index.md) in your index. You can use an open source model like [OpenAI CLIP](https://github.com/openai/CLIP/blob/main/README.md) to vectorize text and images in the same embedding space.|
+| image | vectors <sup>1</sup> | Images can be vectorized in an indexer pipeline, or handled externally for a mathematical representation of image content and then [indexed as vector fields](vector-search-how-to-create-index.md) in your index. You can use [Azure AI Vision multimodal](/azure/ai-services/computer-vision/how-to/image-retrieval) or an open source model like [OpenAI CLIP](https://github.com/openai/CLIP/blob/main/README.md) to vectorize text and images in the same embedding space.|
 <!-- | audio | vectors <sup>1</sup> | Vectorized audio content can be [indexed as vector fields](vector-search-how-to-create-index.md) in your index. Vectorization of audio content often requires intermediate processing that converts audio to text, and then text to vecctors. [Azure AI Speech](/azure/ai-services/speech-service/overview) and [OpenAI Whisper](https://platform.openai.com/docs/guides/speech-to-text) are two examples for this scenario. | 
 | video | vectors <sup>1</sup> | Vectorized video content can be [indexed as vector fields](vector-search-how-to-create-index.md) in your index. Similar to audio, vectorization of video content also requires extra processing, such as breaking up the video into frames or smaller chunks for vectorization. | -->
 
- <sup>1</sup> The generally available functionality of [vector support](vector-search-overview.md) requires that you call other libraries or models for data chunking and vectorization. However, [integrated vectorization](vector-search-integrated-vectorization.md) embeds these steps. For code samples showing both approaches, see [azure-search-vectors repo](https://github.com/Azure/azure-search-vector-samples).
+ <sup>1</sup> Azure AI Search provides [integrated data chunking and vectorization](vector-search-integrated-vectorization.md), but you must take a dependency on indexers and skillsets. If you can't use an indexer, Microsoft's [Semantic Kernel](/semantic-kernel/overview/) or other community offerings can help you with a full stack solution. For code samples showing both approaches, see [azure-search-vectors repo](https://github.com/Azure/azure-search-vector-samples).
 
-<sup>2</sup> [Skills](cognitive-search-working-with-skillsets.md) are built-in support for [AI enrichment](cognitive-search-concept-intro.md). For OCR and Image Analysis, the indexing pipeline makes an internal call to the Azure AI Vision APIs. These skills pass an extracted image to Azure AI for processing, and receive the output as text that's indexed by Azure AI Search.
+<sup>2</sup> [Skills](cognitive-search-working-with-skillsets.md) are built-in support for [applied AI](cognitive-search-concept-intro.md). For OCR and Image Analysis, the indexing pipeline makes an internal call to the Azure AI Vision APIs. These skills pass an extracted image to Azure AI for processing, and receive the output as text that's indexed by Azure AI Search. Skills are also used for integrated data chunking (Text Split skill) and integrated embedding (skills that call Azure AI Vision multimodal, Azure OpenAI, and models in the Azure AI Studio model catalog.)
 
 Vectors provide the best accommodation for dissimilar content (multiple file formats and languages) because content is expressed universally in mathematic representations. Vectors also support similarity search: matching on the coordinates that are most similar to the vector query. Compared to keyword search (or term search) that matches on tokenized terms, similarity search is more nuanced. It's a better choice if there's ambiguity or interpretation requirements in the content or in queries.