Typically (general purpose) build systems make you choose between reproducibility and performance. Please and other related tools aim to change that (however, in my experience they all fail for one reason or another—not because the problem is inherently intractable but rather because these tools are underinvested-in such that they’re infeasible for smaller organizations to operate and larger organizations can afford to roll their own in-house and get the benefits of a tool suited to their specific needs).
I think this “missing middle” is a common pattern in expert tools. The economics are challenging because people quickly move through the skill level where the tools are most useful.