I will say though, for a beginner, the breadth-first approach isn't bad--they don't know a good phaser from a bad phaser, but they know a phaser from a not-phaser; let them learn what a phaser is from a cheap DSP pedal, and if they like the sound, they can buy a real one.
This is more or less how i learned guitar effects, using cheap digital multi units from ~2005-2010; adding a natural language interface to that doesn't have to be bad, though I'd obviously prefer it explain what it's doing and not just presenting an un-investigable final output. Regardless, there is and always will be a market for beginner guitarists, and at the right price point, i could see this being good for them.
I agree a good multi-effect is useful to learn what the different effects actually do, but there are good entry-level multi-effect pedals that are cheaper than that. And this pedal can only have one effect at a time.
Also, it seems there's no preview in their AI playground, so you have to burn tokens and upload the effect to test it, and it may take lots of iterations to get what you want.
So I think this could mainly interest developers who are able to use it as a platform to develop their own effects without going through the AI thing, and beginners who want to be able to use different community effects to test things.