My first guess is that, outside of the rehearsed elements like camera turns, where Ellie goes, etc. that there are genuine bits in the demo where the game plays itself. That last 40 seconds of gameplay in particular looks so perfectly executed, and so beyond what should be feasible when it comes to player input, enemy reactions, context sensitive animations, how smoothly gameplay shifts to a cinematic, etc. that if we got a Jason Schreier BTS with anonymous dev interview I wouldn't be surprised to learn that it was essentially a cinematic.
And it's simply because of the implication. That all of this is entirely avoidable and not scripted. Like how would a cinematic transition happen anywhere else if not in that exact position of the room?