Yeah. I’m convinced the current model is just too confusing. But I really wish there were new interaction patterns that took advantage of low latency speech recognition...