AI Features

Updated on May 21, 2026

The AI features of RadGridView enable natural language interactions with the grid. Users type a text prompt and the grid translates the AI model response into grid operations such as sorting, filtering, grouping, selection, paging, column manipulation, and data export.

The feature introduces a built-in prompt input panel, an extensible command-based API for communicating with an AI model, and a JSON response processor that applies the returned commands to the grid.

Enabling AI Features

To activate the AI prompt panel, set the IsAIEnabled property to True. The grid automatically creates a default GridViewAISettings instance when IsAIEnabled is set to True and no settings object exists. The GridViewAISettings object can be accessed via the AISettings property of RadGridView.

Enabling AI features

xml

<telerik:RadGridView x:Name="gridView' IsAIEnabled="True"/>

C#

this.gridView.IsAIEnabled = true;

The panel that allows AI text input contains a RadSpeechToTextButton that requires to install the following NuGet package: Microsoft.Web.WebView2.

Configuring AI Settings

The AISettings property accepts an instance of GridViewAISettings that controls the prompt input panel behavior.

Configuring AISettings

XAML

<telerik:RadGridView IsAIEnabled="True" ItemsSource="{Binding Items}">
    <telerik:RadGridView.AISettings>
        <gridView:GridViewAISettings InputText=""
                                     SubmitPromptOnSelection="False"
                                     IsSuggestedPromptsVisible="True"
                                     IsRecentPromptsVisible="True"
                                     EnableIsBusy="True"
                                     OpenPromptSuggestionsOnFocus="True" />
    </telerik:RadGridView.AISettings>
</telerik:RadGridView>

The following table lists the available properties of the GridViewAISettings class.

Property	Type	Description
InputText	string	Gets or sets the text in the prompt input field. Supports two-way binding.
SuggestedPrompts	IEnumerable<string>	Gets or sets the collection of suggested prompts displayed in the suggestions dropdown.
RecentPrompts	IEnumerable<string>	Gets or sets the collection of recent prompts. Automatically updated when a prompt is submitted.
SubmitPromptOnSelection	bool	When set to True, selecting a suggested prompt submits the request immediately. Default is False.
IsSuggestedPromptsVisible	bool	Gets or sets whether the suggested prompts section is visible. Default is True.
IsRecentPromptsVisible	bool	Gets or sets whether the recent prompts section is visible. Default is True.
EnableIsBusy	bool	Gets or sets whether the grid shows a busy indicator during prompt processing. Default is True.
OpenPromptSuggestionsOnFocus	bool	Gets or sets whether the prompt suggestions dropdown opens when the input receives focus. Default is True.

The smart AI features of RadGridView work with a special JSON format that is constructed by using a combination between the internal GridView workings and an AI client (like AzureOpenAIClient).

Setting up an AI chat client

C#

IChatClient aiChatClient;

public MainWindow()
{
    this.InitializeComponent();    

    var azureClient = new AzureOpenAIClient(new Uri("azure-AI-resource-address-here"), new Azure.AzureKeyCredential("azure-key-here")).GetChatClient("gpt-5");
    aiChatClient = new ChatClientBuilder(azureClient.AsIChatClient()).UseFunctionInvocation().Build();
}

The aiChatClient from the example above will be used in the HandlePromptRequest to produce the required JSON contents.

Handling Prompt Requests

When the user submits a prompt, the GridView raises the HandlePromptRequest event. Subscribe to this event to send the request to your AI model and set the response JSON.

Handling the prompt request event

C#

this.gridView.HandlePromptRequest += OnHandlePromptRequest;

private async void OnHandlePromptRequest(object sender, GridViewPromptRequestCommandEventArgs e)
{
	 var request = JsonSerializer.Deserialize<GridAIRequest>(e.RequestJson);
	 var options = new ChatOptions();
	 ChatOptionsExtensions.AddGridChatTools(options, request.Columns);
	 List<ChatMessage> messages = request.Contents.Select(m => new ChatMessage(ChatRole.User, m.Text)).ToList();
	 ChatResponse completion = await this.chatClient1.GetResponseAsync(messages, options);
	 var gridResponse = completion.ExtractGridResponse();
	 var prop = new JsonSerializerOptions() { PropertyNamingPolicy = JsonNamingPolicy.CamelCase };
	 var response = JsonSerializer.Serialize(gridResponse, prop);

	 e.ResponseJson = response;
}

The example above requires to install the following NuGet packages:

• Azure.AI.OpenAI
• Microsoft.Extensions.AI.OpenAI
• Telerik.AI.SmartComponents.Extensions

The GridViewPromptRequestCommandEventArgs class provides the following members:

Member	Type	Description
Prompt	string	The user's text prompt.
RequestJson	string	The JSON payload with grid metadata (columns and content) sent to the AI model.
ResponseJson	string	Set this property with the AI model's JSON response. The grid processes the commands when the value is set.
HasError	bool	Set to True to indicate an error during processing.

Handling Prompt Request Cancellation

To handle cancellation of an in-flight prompt request, subscribe to the CancelPromptRequest event.

Handling cancellation

C#

this.gridView.CancelPromptRequest += OnCancelPromptRequest;

private void OnCancelPromptRequest(object sender, EventArgs e)
{
    // Cancel the ongoing AI request.
}