Start with a process, not a tool.
Pick a task that happens frequently and has a clear trigger and outcome. Map what comes in, what happens next and where people make decisions. A small, well-defined workflow is easier to evaluate than an entire department.
Measure the work today.
Record how often it runs, how long it takes and how often an exception occurs. Use your own numbers to estimate the opportunity. Time saved is useful, but faster response and better consistency may matter just as much.
Keep the right decisions human.
Separate repeatable actions from judgement. An automation may prepare information, while a person approves the next step. That boundary should be part of the design, not an afterthought.
Run a focused pilot.
Agree on a measurable outcome and review real examples together. Expand only after the first workflow is understood, tested and useful.