Before you decide which tool to buy, ask whether the work you’re about to automate makes sense as it is.
Most of the AI conversations I hear right now start in the same place: which tool should we buy?
It’s an understandable question. The pressure to do something is real, and the promises are hard to ignore. But it skips the question I think matters more: does the work we’re about to automate make sense as it is? If a process is already unclear, technology won’t clean it up. It will just move the confusion faster.
What I see instead is companies treating the rollout as a training problem. People learn how to use the tool, and the project is called done. What never gets decided is when the tool should be used, who reviews what it produces, which calls still need a person and who is accountable when something goes wrong. Those aren’t technology questions. They’re questions about how the work is designed.
When they go unanswered, people end up doing two jobs. The tool handles part of the task, but someone still has to check it, fix it, deal with the exceptions and explain the result, and that work almost never makes it into the efficiency estimate. Roles get fuzzier, too. Employees aren’t sure which decisions to trust the system with. Managers aren’t sure how to evaluate work that is part person and part system. And accountability slips, because everyone assumes someone else is checking.
I’ve seen a version of this with technology rollouts long before AI. A new system goes live, and people keep their spreadsheets and email threads running right beside it. It gets called resistance, and sometimes it is. But often that old workaround is holding information or judgment the new system never accounted for. The company ends up paying for the new tool while maintaining the old process underneath it, and instead of removing work, the rollout adds another layer.
So before any rollout, I want leadership to be able to answer these questions:
- What problem are we solving?
- Which part of the work will change?
- Which part still requires human judgment?
- Who is accountable for the final result?
- What will employees stop doing?
- How will roles and performance expectations change?
- What information will the tool use?
- Who reviews errors or unexpected outcomes?
- How will we know whether the change worked?
And I want the people who actually do the work to test the new process before it goes company-wide. They will see what doesn’t hold up long before a project plan will.