Package com.smartgwt.client.ai

Class AIMessage

java.lang.Object
com.smartgwt.client.core.JsObject
com.smartgwt.client.core.DataClass
com.smartgwt.client.ai.AIMessage
All Implemented Interfaces:
HasHandlers
Direct Known Subclasses:
AIMessageWithSource
public class AIMessage extends DataClass
An individual message in the list of AIRequest.messages of an AI request.

  • Constructor Details

    • AIMessage

      public AIMessage()

    • AIMessage

      public AIMessage(JavaScriptObject jsObj)

  • Method Details

    • getOrCreateRef

      public static AIMessage getOrCreateRef(JavaScriptObject jsObj)

    • setContent

      public AIMessage setContent(Object content)
      The data for the message. If the type of this message is "jpegImage" or "pngImage", this will be the base64-encoded image data.
      Parameters:
      content - New content value. Default value is null
      Returns:
      AIMessage instance, for chaining setter calls

    • getContent

      public Object getContent()
      The data for the message. If the type of this message is "jpegImage" or "pngImage", this will be the base64-encoded image data.
      Returns:
      Current content value. Default value is null

    • setImageDetail

      public AIMessage setImageDetail(String imageDetail)
      For a message whose type is "jpegImage" or "pngImage", how finely the AI should examine the image: "low", "high" or "auto". Only meaningful for engines that expose such a setting; ignored by the others.

      When unset, the engine's own default applies, which for most engines downsamples the image substantially. Set "high" when small text in the image must be read - the text in a sketch or wireframe, for instance - and expect a correspondingly higher token cost.

      Parameters:
      imageDetail - New imageDetail value. Default value is null
      Returns:
      AIMessage instance, for chaining setter calls

    • getImageDetail

      public String getImageDetail()
      For a message whose type is "jpegImage" or "pngImage", how finely the AI should examine the image: "low", "high" or "auto". Only meaningful for engines that expose such a setting; ignored by the others.

      When unset, the engine's own default applies, which for most engines downsamples the image substantially. Set "high" when small text in the image must be read - the text in a sketch or wireframe, for instance - and expect a correspondingly higher token cost.

      Returns:
      Current imageDetail value. Default value is null

    • setType

      public AIMessage setType(AIContentType type)
      The type of data.
      Parameters:
      type - New type value. Default value is "text"
      Returns:
      AIMessage instance, for chaining setter calls

    • getType

      public AIContentType getType()
      The type of data.
      Returns:
      Current type value. Default value is "text"