Updating WinML docs to focus on self-contained (#1147)

* Updating WinML to target the WASDK component package

* Updates to overview

* Review updates

* Apply suggestions from code review

Co-authored-by: Copilot <[email protected]>

---------

Co-authored-by: Copilot <[email protected]>

Author: andrewleader

docs/new-windows-ml/distributing-your-app.md

@@ -1,7 +1,7 @@
 ---
 title: Deploy your app that uses Windows ML
 description: Learn how to deploy your app that uses Windows Machine Learning (ML).
-ms.date: 03/10/2026
+ms.date: 03/20/2026
 ms.topic: concept-article
 ---
 
@@ -11,11 +11,11 @@ Windows ML is deployed like any other Windows App SDK component, supporting both
 
 ## Framework-dependent
 
-Your app depends on the Windows App SDK runtime and/or framework package being present on the target machine. Framework-dependent deployment is the default deployment mode of the Windows App SDK for its efficient use of machine resources and serviceability. See [Deployment architecture and overview for framework-dependent apps](/windows/apps/windows-app-sdk/deployment-architecture) for more details.
+Your app depends on the Windows App SDK runtime and/or framework package being present on the target machine. This minimizes the dependencies your app has to carry (by using shared system-wide copies of Windows App SDK runtime), and enables evergreen servicing without you needing to release an update to your app. When targeting the main [`Microsoft.WindowsAppSDK` NuGet package](https://www.nuget.org/packages/Microsoft.WindowsAppSDK), framework-dependent deployment is the default deployment mode. See [Deployment architecture and overview for framework-dependent apps](/windows/apps/windows-app-sdk/deployment-architecture) for more details.
 
 ## Self-contained
 
-Your app carries the Windows App SDK dependencies with it. See [Deployment guide for self-contained apps](/windows/apps/package-and-deploy/self-contained-deploy/deploy-self-contained-apps) for more details.
+Your app carries the Windows ML dependencies with it. When your app uses individual Windows App SDK component NuGet packages, like the [`Microsoft.WindowsAppSDK.ML` NuGet package](https://www.nuget.org/packages/Microsoft.WindowsAppSDK.ML) for Windows ML, self-contained deployment is the default deployment mode. See [Deployment guide for self-contained apps](/windows/apps/package-and-deploy/self-contained-deploy/deploy-self-contained-apps) for more details.
 
 In self-contained mode, ONNX Runtime binaries are deployed alongside your application:
 

docs/new-windows-ml/get-started.md

@@ -1,15 +1,15 @@
 ---
 title: Get started with Windows ML
 description: Learn how to use Windows ML to download and register AI execution providers for hardware-optimized inference.
-ms.date: 03/16/2026
+ms.date: 03/20/2026
 ms.topic: how-to
 ---
 
 # Get started with Windows ML
 
-This topic shows you how to install and use Windows ML to discover, download, and register execution providers (EPs) for use with the ONNX Runtime shipped with Windows ML. Windows ML handles the complexity of package management and hardware selection, automatically downloading the latest execution providers compatible with your device's hardware.
+This topic shows you how to install and use Windows ML to discover, download, and register execution providers (EPs) for use with the ONNX Runtime shipped with Windows ML. Windows ML handles the complexity of package management and hardware selection, allowing you to download the latest execution providers compatible with your users' hardware.
 
-If you're not already familiar with the ONNX Runtime, we suggest reading the [ONNX Runtime docs](https://onnxruntime.ai/docs/). In short, Windows ML provides a shared Windows-wide copy of the ONNX Runtime, plus the ability to dynamically download execution providers (EPs).
+If you're not already familiar with the ONNX Runtime, we suggest reading the [ONNX Runtime docs](https://onnxruntime.ai/docs/). In short, Windows ML provides a copy of the ONNX Runtime, plus the ability to dynamically download execution providers (EPs).
 
 ## Prerequisites
 
@@ -43,7 +43,11 @@ Python versions 3.10 to 3.13, on x64 and ARM64 devices.
 
 Windows ML is included in [Windows App SDK 1.8.1 or greater](/windows/apps/windows-app-sdk/stable-channel).
 
-See [use the Windows App SDK in an existing project](/windows/apps/windows-app-sdk/use-windows-app-sdk-in-existing-project) for how to add the Windows App SDK to your project, or if you're already using Windows App SDK, update your packages.
+The easiest way to use Windows ML is to install the [`Microsoft.WindowsAppSDK.ML` NuGet package](https://www.nuget.org/packages/Microsoft.WindowsAppSDK.ML), which uses self-contained deployment by default. See [Deploy your app](./distributing-your-app.md) to learn more about deployment options.
+
+```bash
+dotnet add package Microsoft.WindowsAppSDK.ML
+```
 
 ### [C++/WinRT](#tab/cppwinrt)
 
@@ -53,7 +57,7 @@ See [use the Windows App SDK in an existing project](/windows/apps/windows-app-s
 
 ### [C/C++](#tab/c)
 
-Windows ML C APIs are included in the [`Microsoft.WindowsAppSDK.ML` NuGet package](https://www.nuget.org/packages/Microsoft.WindowsAppSDK.ML) version 1.8.6 or greater. The NuGet package ships CMake config files under `build/cmake/` that are directly consumable. The `@WINML_VERSION@` token is resolved at NuGet build time, so no additional processing is required.
+Windows ML C APIs are included in the [`Microsoft.WindowsAppSDK.ML` NuGet package](https://www.nuget.org/packages/Microsoft.WindowsAppSDK.ML) version `1.8.2141` or greater. The NuGet package ships CMake config files under `build/cmake/` that are directly consumable. The `@WINML_VERSION@` token is resolved at NuGet build time, so no additional processing is required.
 
 To use the NuGet package...
 

docs/new-windows-ml/migrate-to-windows-ml.md

@@ -1,7 +1,7 @@
 ---
 title: Migrate from standalone ONNX Runtime to Windows ML's ONNX Runtime
 description: Learn how to migrate from using the standalone ONNX Runtime to using the ONNX Runtime included in Windows Machine Learning (ML) for hardware-optimized inference.
-ms.date: 08/14/2025
+ms.date: 03/20/2026
 ms.topic: how-to
 ---
 
@@ -13,9 +13,9 @@ This guide explains how to migrate from using the [standalone ONNX Runtime](http
 
 - **Smaller app download / install size** - Your app doesn't need to distribute large EPs and the ONNX Runtime
   - **EPs are dynamically downloaded via Windows ML**, so that you don't have to bundle them with your app
-  - **The ONNX Runtime in Windows ML is a shared system-wide copy**, so your app doesn't have to bundle it with your app
-- **Dynamically uses latest EPs** - Automatically downloads and manages the latest compatible hardware-specific execution providers, without requiring your app to update
-- **Dynamically uses latest ONNX Runtime** - Automatically updates the ONNX Runtime without requiring your app to update. See the [ONNX versions](./onnx-versions.md) docs for more info
+  - **Optionally use a shared system-wide ONNX Runtime**, so your app doesn't have to bundle it with your app
+- **Evergreen EPs** - Automatically updates to the latest compatible hardware-specific execution providers, without requiring your app to update
+- **Optional evergreen ONNX Runtime** - By using framework-dependent deployment, your app can automatically receive updates to the ONNX Runtime without requiring your app to update. See the [ONNX versions](./onnx-versions.md) docs for more info
 
 ## System requirements for Windows ML
 

docs/new-windows-ml/onnx-versions.md

@@ -1,13 +1,13 @@
 ---
 title: ONNX Runtime versions shipped in Windows ML
 description: Understand which versions of the ONNX Runtime were shipped in which versions of Windows ML.
-ms.date: 03/19/2026
+ms.date: 03/20/2026
 ms.topic: concept-article
 ---
 
 # ONNX Runtime versions shipped in Windows ML
 
-Each Windows App SDK release includes Windows ML, which includes a copy of the ONNX Runtime, so that your app can depend on a shared system-wide copy of the ONNX Runtime rather than distributing your own copy.
+Each Windows App SDK release includes Windows ML, which includes a copy of the ONNX Runtime, so that your app can depend on a shared system-wide copy of the ONNX Runtime rather than distributing your own copy (if you choose framework-dependent deployment). See [Deploy your app](./distributing-your-app.md) for more details.
 
 ## Versions of ONNX Runtime in Windows ML
 

docs/new-windows-ml/overview.md

@@ -2,7 +2,7 @@
 title: What is Windows ML?
 description: Learn how Windows Machine Learning (ML) helps your Windows apps run AI models locally.
 ms.topic: article
-ms.date: 11/18/2025
+ms.date: 3/20/2026
 ---
 
 # What is Windows ML?
@@ -15,9 +15,9 @@ If you're not already familiar with the ONNX Runtime, we suggest reading the [ON
 
 ## Key benefits
 
-- **Dynamically get latest EPs** - Automatically downloads and manages the latest hardware-specific execution providers
-- **Shared [ONNX Runtime](https://onnxruntime.ai/docs/)** - Uses system-wide runtime instead of bundling your own, reducing app size
-- **Smaller downloads/installs** - No need to carry large EPs and the ONNX Runtime in your app
+- **Shared [ONNX Runtime](https://onnxruntime.ai/docs/)** - Optional system-wide runtime instead of bundling your own, reducing app size
+- **Dynamically get latest EPs** - Download the latest hardware-specific execution providers with one API call
+- **Smaller app downloads/installs** - No need to carry large EPs and the ONNX Runtime in your app
 - **Broad hardware support** - Runs on Windows PCs (x64 and ARM64) and Windows Server with any hardware configuration
 
 ## System requirements
@@ -39,14 +39,14 @@ You can [see the list of EPs that Windows ML supports here](./supported-executio
 
 Windows ML includes a copy of the [ONNX Runtime](https://onnxruntime.ai/) and allows you to dynamically download vendor-specific **execution providers** (EPs), so your model inference can be optimized across the wide variety of CPUs, GPUs, and NPUs in the Windows ecosystem.
 
-### Automatic deployment
+Execution provider acquisition:
 
-1. **App installation** - Windows App SDK bootstrapper initializes Windows ML
-2. **Hardware detection** - Runtime identifies available processors  
-3. **EP download** - Automatically downloads optimal execution providers
-4. **Ready to run** - Your app can immediately use AI models
+1. **Hardware detection** - Windows ML identifies compatible EPs
+2. **EP download** - Call APIs to download compatible EPs
+3. **Ready to run** - Your app can accelerate AI models with EPs
 
 This eliminates the need to:
+
 - Bundle execution providers for specific hardware vendors
 - Create separate app builds for different execution providers
 - Handle execution provider updates manually