Discovery and browsing are not good on the assistant interface but I'd argue that's a constraint of voice vs visual. On a desktop/mobile, the screen holds the state so you can go back to previous entries, etc. Over voice, you mind holds the state which scales much much worse.