If you can automate efficiently enough, you can build far finer structures.
For example look at how detailed the structure and weld arrangement is for modern cars, vs. back when robots only just started to take care of the frame welding on the assembly line.
Or how optical HDMI cables are affordable because they use fully automated UV-cure-resin-glued fiber alignment straight from the cable end into the optoelectronic chips, without needing optical connectors or any other human-labor to get the light path connected up. That's how they manage to do it the conceptually easiest way: amplifier->laser->fiber->photodiode->amplifier, and repeat for the 4 high-speed pairs. Also handing the low speed communication channel separately with just normal wires as signal degradation isn't an issue for that.
Or for example 3d printer infill: that's something no one would do manually in such a way, but if it's just automated it's quite desirable/efficient.
App rental e-scooters: they rely on automation to organize even when parked "pretty much anywhere they're not gonna block traffic", and as such become relevant for even short trips.
If you have an unsupervised robot that lays bricks for you to build up a house, you can get away with smaller bricks (and thus a lighter/cheaper machine needing a smaller crane to lift up/out of higher floors), than if you need a human to supervise it.
Smaller machine if slower means more machines, meaning cheaper production of the machines due to scale.
Auto-feeders for nail guns in construction means more smaller nails as placing 3 in a row takes barely longer than just 2.
Especially if the nail gun could, say, run like an optical mouse and automatically trigger at a configured spacing while dragged along a surface with the trigger held down.
For example look at how detailed the structure and weld arrangement is for modern cars, vs. back when robots only just started to take care of the frame welding on the assembly line.
Or how optical HDMI cables are affordable because they use fully automated UV-cure-resin-glued fiber alignment straight from the cable end into the optoelectronic chips, without needing optical connectors or any other human-labor to get the light path connected up. That's how they manage to do it the conceptually easiest way: amplifier->laser->fiber->photodiode->amplifier, and repeat for the 4 high-speed pairs. Also handing the low speed communication channel separately with just normal wires as signal degradation isn't an issue for that.
Or for example 3d printer infill: that's something no one would do manually in such a way, but if it's just automated it's quite desirable/efficient.
App rental e-scooters: they rely on automation to organize even when parked "pretty much anywhere they're not gonna block traffic", and as such become relevant for even short trips.
If you have an unsupervised robot that lays bricks for you to build up a house, you can get away with smaller bricks (and thus a lighter/cheaper machine needing a smaller crane to lift up/out of higher floors), than if you need a human to supervise it.
Smaller machine if slower means more machines, meaning cheaper production of the machines due to scale.
Auto-feeders for nail guns in construction means more smaller nails as placing 3 in a row takes barely longer than just 2. Especially if the nail gun could, say, run like an optical mouse and automatically trigger at a configured spacing while dragged along a surface with the trigger held down.