They haven’t been truly novel if you want to say that, for example, the Lisa was covering Xerox PARC ideas but I think you’d have to ignore a lot of significant work to say they didn’t substantially innovate in GUIs, personal assistants and handwriting recognition (Newton), touchscreen behavior (iPhone), etc.

The key thing is that they tend not to ship things which aren’t mature enough to be useful (Vision Pro and Apple Intelligence being cautionary tales about why) and voice assistants just aren’t doing a whole lot for anyone. Google and Amazon have been struggling to find a market, too, and it’s rare to find someone who uses those dramatically more than Apple users do Siri. I think most of the big advances need something close to AGI: if you can’t describe something in a short command, it’s usually much faster to use a device with a screen and a lot of the useful tasks need a level of security and reliability which requires actual reasoning to deliver.