Right but the visual experience isn't as much a differentiator as the "does this answer my question" experience, which relies on fairly sota methods in NLP, clustering, etc.