for a lot of teams that do pair programming etc and pairing with agents - this makes sense.

hell dare I say better than the pull-request model - since you don't have to wait on people to review the code and deal with the nonsense of merge-trains.