With a few hundred highly trained consultants spread through major markets and working directly with contractors with decent jobflow to train them, it could work.

Glulam beams might be manufacturable on site—and it might even be better, since you can make very long glulam beams, but long beams are more expensive to ship.

In some places in the world (like Pakistan), it is common for upper middle class people to have custom cabinetry in their homes. Usually, during house construction, the carpentry team shows up with the necessary cutting machines. Usually a back of pickup van is sufficient.

Everything is cut from full sized boards as needed on site and installed right there.

It works because wood working machinery is fairly cheap, not heavy and works off standard electric power.

When cabinetry (custom or otherwise) is manufactured, it's done with large CNC machines and then assembled.

Unless you were doing a _lot_ of cabinets, the labor cost of an idle cabinet maker overwhelms the shipping cost.

On-site, you're paying the cabinet maker to sand the wood.

https://www.youtube.com/watch?v=WGEF1T3omec

Another of the companies mentioned, Facit, has a lot more videos and information about their build process:

https://www.facit-homes.com/building-process

I visited Re:frame Systems[1] in Boston last month and it was pretty cool to see their system which was apparently primarily comprised of a series of magnetic stops on an angled work-surface tended by a giant robotic arm. Cut to length lumber would be fed in on batched trolleys and the arm would compensate for non-square lumber while pinning individual pieces in place then finally nailgun them together in to a completed structural frame. These frames then moved flatpack-style to a second site in the factory for assembly then manual fit-out. I was quite surprised at how slow the arm was, but it didn't really matter because a human would likely take longer by the time they went up a ladder a few times, had a break, and took time off for weather interruptions. They had a pretty impressive team drawing experience from most of the larger US robotics applications and were moving from single level to double level housing, with more elaborate structures planned in future.

[1] https://www.reframe.systems/