I'm guessing it is just referring to reliability of rockets in general. Reliability should be improved and being able to fly more rockets by being "fully and rapidly reusable" is the right approach. Planes would not have been able to become as safe as they are if we had to build a new one for each flight.
Soyuz is the only rocket that has flown much more than 100 flights, and it has a reliability of 97%.
Several other rockets, including Falcon 9, have flown enough successful flights that statistically their reliability is unlikely to be less than 99%, but we cannot know that it's any higher than that.
NASA figures that the failure rate of a Dragon flight is about 1 in 500, but that's based on analysis rather than statistics and is likely quite wrong. Note that the highest chance of failure in their analysis is a micrometeorite collision rather than a rocket failure.
>Soyuz is the only rocket that has flown much more than 100 flights
Falcon 9 has had 140 flights of which 138 were successful, but that includes 1.0 and 1.1 which were significantly different early designs. It's had 120 flights for F9 FT, all successful (though over 111 flights ago one was destroyed in preflight testing). It's arguably the safest rocket system ever made [0].
That said, Starship should easily surpass it. It has more margin to work with and will run at a much higher cadence with much more opportunity for refinement and testing.
Its extremely high flight rate makes it likely that its reliability will be closer to planes than rockets eventually.