An AI project can require more supervision than leadership expects. The promise is fewer clerical errors and less routine work. But someone may still have to check the results, correct mistakes and keep the work moving.
Those responsibilities need to be understood before the investment is made. Otherwise, a useful improvement can end up disappointing everyone because the expectation was wrong.
I think of AI like a very fast sports car. It may take a corner much faster than a pickup, but it still has limits. Knowing where to brake is part of being able to use its speed. Operations leaders need that same understanding before handing over work.
Start with a problem the operation actually has. Repeated data entry and moving information between systems are opportunities I keep coming back to. Then look closely at the proposed process, including the situations in which it could struggle.
Suppose AI is being considered for transferring information between two systems. Comparing their formats is a useful start. What happens when a required field is missing, two records conflict, or an unfamiliar format arrives? Someone needs to establish what the process should do in those circumstances and where a person has to become involved.
Did the delivery quantity make it across?
If the process still needs someone to check every result, that work has a cost. Include the time spent reviewing and correcting AI’s work when calculating the savings.
That means leadership has to spend time understanding the technology as well as listening to its promised benefits. Ask about shortcomings. Look at what has gone wrong in comparable uses. Work through the consequences of an error in the specific operation being considered.
I am optimistic about what AI can do for logistics. There is a lot of repetitive work it could help remove, freeing people to deal with bigger problems. Understanding the limits is how we put that capability to work with expectations the operation can meet.
