Hot take after talking to people about AnveVoice: website voice agents should not be everywhere.
I don’t think generic browsing needs voice. I don’t think checkout should be interrupted. I don’t think a simple UI should be replaced just because AI exists.
The use case only feels strong when a visitor already has intent and friction is blocking the next step.
Examples:
In those cases, voice could answer the one question that decides whether someone leaves or moves forward.
Curious how others think about this: if your website had a voice AI layer, where would you put it first?
A one-week feature took two months, mostly spent keeping three systems in sync