"Automate it" is not, by itself, useful advice. Plenty of processes are manual because manual is genuinely the right call at that organization's current scale. Automating them just adds a layer of software to maintain without removing any real friction.
Start with where the friction actually is
Before we design anything, we want to see the process as it's actually run today, not the version described in a policy document. That usually means sitting with the people doing the work, not just the people managing them.
The processes worth automating are usually the ones with a repeatable, well-understood shape and a clear cost to doing them manually: hours lost, errors that cascade, or a bottleneck that caps how much the business can take on.
Redesign before you automate
Automating a bad process just makes the bad process faster. Part of our job is pointing out when the right move is to redesign the workflow first, even if that's a less exciting deliverable than a new piece of software.