They could revert to the pre-LLM craze where Alexa was actually able to answer prompts in a reasonable amount of time. Alexa is a glorified egg timer for nearly all users - it's super convenient to use a voice prompt to start/stop timers while cooking and being able to play music on voice demand is kinda alright. It genuinely doesn't need to answer hard questions.

Alexa had already been using other models like Claude depending on the question, and the French version for example uses Mistral.

Basic pre-LLM speech-to-text and a passthrough to duckduckgo would beat their current product, so...

Alexa+ was their LLM based release and to my knowledge it was an utter disaster and failure of product development.

Latency is the #1 UX killer on voice apps, and the latency was terrible compared to Alexa. I used it for like 15 minutes and reverted.

All anyone was talking about was how much worse it was for common tasks.