I'd try to keep the video and PDF production connected from the beginning instead of treating them as two separate projects.
For example, build one structured source document first: the actual course outline, key points, examples and scripts. Then use that as the source for both the PDF and the video scripts.
AI can handle a lot of the repetitive work, but I'd still keep a human review stage for the videos. Especially with AI voice/avatar content, small mistakes in pacing, emphasis or visuals can make the whole thing feel very generic.
If speed is the priority, I'd optimize the workflow around producing a good first draft quickly and then spending the human time on the parts the viewer actually notices.