A college computer lab had a simple rule. Before you could ask a teaching assistant for help, you had to explain your problem to a teddy bear on a desk outside the office. A good share of the students never made it past the bear. Saying the problem out loud, with every assumption and shortcut, was enough to find the bug.
I think of that bear every time someone tells me AI makes their team smarter. Much of the value of a chatbot is the bear effect: you are forced to write down what you actually want. Teams that skip that step get confident answers to the wrong question. So before you buy another assistant, try the bear. If you cannot explain the problem to a toy, no model will save you.