Lima (1) is a project that packages Linux distros for MacOS and executes them via qemu in the backend. Maybe you could solve your problem by launching one of their vms and inspecting the command line it generates. You might find an option you were missing.
You are getting a lot of troll replies, but it is actually interesting to think about.
The waste is uncaptured value. It is some part of your software business's domain that is just too hard, expensive, or requiring physical intervention to encode the process in their system. So the business never chooses to build that feature. This leaves some part of your business's problem domain unsolved. Potentially someone else smaller could come in and try and solve that problem and capitalize on that wasted value.
I think human programming is not untapped at all. This is describing every line of business application in existence. A great example of both the power and limitations of this is phone trees for customer support.
A product team and dev team encode business knowledge and flows into code and leverage a human to make judgement calls when necessary. The outcome is a program that can either be used by skilled workers to multiply their output or allow unskilled workers to perform tasks that would have formerly required a skilled worker to accomplish.
There are already (arguably) optimized flows and design patterns for application UX. Companies have already spent years trying to build and optimize this "human programming". Dev teams have developed many DSLs to make it easier to encode business logic into their applications more quickly.
I am not saying line of business applications are good or near some optimal final form, but to call "human programming" untapped is taking a very narrow view of the definition.