The marketing pages, the whole pitch was that we have a native grounds up multi modal that processes multi modal inputs as a whole unlike other approaches where

The marketing pages, the whole pitch was that we have a native grounds up multi modal that processes multi modal inputs as a whole unlike other approaches where multiple different models of different modalities are stiched together. With all that emphasized, the duck demo was jaw dropping. It felt like now computer systems really have eyes and ears. A computer system is responding and commenting on what it is seeing. That's just really really revolutionary. But then... the fine print somewhere. Which should have been first slide on the video in bold headings staying for a few seconds.