feat(ai): make prompt multimodal for generateImage/generateVideo, pass text through verbatim view more