"we have improved the buffer"
When you read claims like that you have to ask *what* is better.
In one man's eye cheaper is better

.
One "flaw" in discrete buffers is the slight gain loss (roughly ~0.075dB for a BJT buffer).
So if you have two buffers in the effect you might get 0.15dB. So when you have a number of
effects in cascade the loss adds up, perhaps to 0.5dB to 1dB. Usually it's not noticed and not a problem.
The JFET buffers can have quite a bit more loss.
One fix for this is to use a buffer with feedback so the gain is 1.
I can't find a good example schematic but you would use a circuit like this with C2 and R4 removed,
then the emitter of the first transistor connects to the collector of the second transistor.
(You can use whatever biasing scheme you like on the base of Q1, like a single resistor
to Vref.)

JFET version:

I'm not saying that's what Boss use. All I'm saying is there's is a small problem and that's one way to fix it with a discrete design. Another common method is to replace the emitter resistor of the one-transistor buffer circuit with a current source.