Video applications even for tech roles are Ok with me. I got a job that way once (at least that was round 1).
But it sounded from the pitch that you are using AI to make the decision or at least kick a lot of people out at a stage, and that doesn’t sound great for either party.
My fear is this will just bias against some portion of the population. Maybe against men, maybe against certain races, or against older people. Because they didn’t on average have the same micro facial expressions, vocal variety or hand expressions as what you are optimising for.