Every. Damn. Time. I've fought and lost this battle at so many companies with directors and executives who were genuinely completely delusional about how much of a human process could be automated.

It's always the tasks that are easy and don't take up much time. If you end up solving one of the genuinely hard problems then you should just pivot to packaging and selling that as your business.

Check out story of scalefactor.

It might not be optimal, a human tax professional might be more creative and save you money, but it should always produce correct forms.

Many of the most reliable systems you can think of work this way, from the mail, to taxes, to factories, to autopilot systems. The key to building them is to be intentional about what you're doing and especially avoid blaming the humans for the system's failures.