More voice assistants

In Seattle, I still have yet to hear anybody talking to an AI assistant in public. It doesn’t seem like it at least, since AI chats are really assistant-coded: one input and wait for a response output with no active listening cues really.

I’m a big fan of the newly launched Raycast iOS app which comes with a great UX for speech to text transcription, even directly from a widget. It’s a hold to transcribe + send interface, and as you release, the LLM immediately starts responding in text. Also, major shoutout to their amazing team who has fixed the only two nits I’ve ever had with their products in under a week both times!

There are and always have been the loud phone calls in public, including embarrassing ones. So I felt like it might have been a UX and reliability problem causing the lack of adoption in at least speech to some type of LLM response for the past few years. But still today, even after all these new great interfaces and reduced latencies, I can’t tell that there are many people in public asking their silly questions out loud.

Grok voice handles great now, even on long calls, Perplexity voice realtime search and source-citing is cool, and the Raycast audio to text with a text response is also great. So I am expecting quite a lot more adoption this year, and expect to hear lots of those silly overly-specific questions in public.

I think I might start doing it a bit. Not obnoxiously, just like a chill phone call type of thing. Typing on the phone all the time is getting tiresome, and voicing the whole thing out start to finish in one go helps me practice my conciseness. It’s just a little embarrassing is all.