It doesn't look all that different to me. Seems like it's just saving you mess around with assigning function inputs and outputs to shell variables. Otherwise it just looks like piping stuff around between functions.
You can write many of the examples we provide in bash using tee and tee >(process) syntax when you pipe data into multipipe blocks. To collect the data from multipipe blocks you need to construct Unix domain named pipes and use them in exactly the right order. It quickly gets complicated and ugly. This is our fourth stab at the problem. The earlier ones generated bash scripts, which looked awful and were unreliable.
I think my main reason for posting was to suggest showing the equivalent bash so it was easier to see what the tool did. I just threw in the "looks potentially great" as a little sweetener. ;)