Of course research outputs should include functioning code and data for reproducibility whenever possible, or even easily reusable software where it’s feasible. But it doesn’t make sense to generally expect from academia to build software that engineers can just plug directly into their systems.