* There are several small typos on the page ("How does it works?") that distract from its content.
* The testimonials strike me as all having the same writer's voice.
* What appears to set "Audrey" aside from, say, Alexa is the promise that you can take a moment to train the AI when it misunderstands you. That's a great feature; I can tell Alexa that she was wrong (via the app), but that's tedious and may take months to see my feedback enacted. Immediately linking what I said to the desired outcome for future use would be ideal. I'd like to see an example of what that looks like for this app, even if it's just a mock-up.