To ask or not to ask; that is the question.
Considerations for choosing a command or conversational interaction model when building for voice technology
If you’ve ever wanted to try using a voice technology feature, you may have wondered how it would respond to you. For those who build them, the consideration can also be if it responds — at all. If it doesn’t, it’s generally known as a command interface. If it does, it’s seen as a conversational interface.
Spoiler alert: if you talk to it and it responds in any way, it’s a conversational interface.
Let’s take some examples. Apple’s virtual assistant Siri exhibits a command model when you give a specific enough command like, “open Wordle”:
Siri: (says nothing; just executes the action — opens the Wordle app)
Obvious or not, no verbal response is needed. However, a command that involves a possibly irreversible action such as, “Send a message to Simone with WhatsApp saying ‘I love you’” kicks off a short and efficient but very much necessary, conversation:
Siri: Send it?
(with the visual preview of the message compiled and the buttons to send it manually available if you decide to do so)
Me: Yes.
Siri: Done.
