Create new class MultimodalUtterance?

[NoB0]

Should we create a new type of instance to handle multimodality (e.g., images, buttons)?

[kbalog]

Yes, that's a good idea.

[NoB0]

After discussion, the MultimodalUtterance class is not needed. Instead, we can use metadata at the Utterance level.

[kbalog]

Where did that discussion take place? :)
While functionally equivalent, having a separate class would make it more clear for someone reading the code when a multimodal response is expected.