Comment by kevintouati

Interesting angle. Does the search use semantic embeddings so it can surface clips by concept rather than filename/metadata? If it nails the retrieval part, that could be the real differentiator.