-
Notifications
You must be signed in to change notification settings - Fork 12
Description
Current Limitation
Currently, users need to manually type out the syntax to add multi-modal inputs - such as file, image, and audio documents - into the expression editor, when writing prompts for model providers:
${<ai:ImageDocument>{content: "https://ballerina.io/img/branding/ballerina_logo_dgrey_png.png"}}
${<ai:FileDocument>{content: "https://ballerina.io/pdf/ballerina_docs.pdf"}}
${<ai:AudioDocument>{content: "https://ballerina.io/mp3/branding/tech_talk.mp3"}}This approach is not user-friendly for low-code users, as it requires prior knowledge of the syntax and document types.
Suggested Improvement
Enhance the expression helper pane to allow users to easily insert multi-modal document objects into the expression editor.
The helper pane currently supports inserting items like Inputs, Variables, Configurables, and Functions (as shown below).
We can introduce a new section — for example, Documents — that lets users conveniently add ai:ImageDocument, ai:FileDocument, or ai:AudioDocument objects when editing an expression field of type ai:Prompt.
Version
1.5.0