A messy process can become faster with automation, but still produce the same mistakes. Before introducing a system, it is worth observing how the work is done and why some steps are repeated.
Consider a request that comes by email, is copied into a sheet and then transcribed into another program. Automating copies can reduce manual labor. But if an essential fact is missing or no one knows who must approve the request, the main problem remains organizational.
I propose to describe the path from the entrance to the conclusion, identifying the people responsible, the necessary information and expected result of each phase. Steps without a clear function can be deleted or merged. Exceptions must be recognized, because they often take up more time than the ordinary route.
Only after this review should you choose what to automate. Repetitive activities with stable rules lend themselves to a different assessment than those requiring interpretation. An automated process should also allow you to identify errors and resume work when the system stops.
A limited trial allows you to compare times, corrections, and requests for assistance before and after the change. The digital transformation of SMEs, studied by the OECD, offers the general context; the simplification proposed here is an operational method to be adapted to the activity. The useful result is a more understandable path for those who perform it and for those who await its outcome.