Instead of enabling knowledge answering on the node you now do it on the capture information cards like the Raw Input card Robert has in the video.