"Do you want to switch or not?"

“Do you want to switch or not?”

Late last year a doorstep salesman went viral because he opened his pitch exactly like that. No introduction, no explanation, no context. Just the question.

The internet did its thing. People were outraged, people laughed, and people went on about it far too long and far too earnestly.

Even so, I think the man was well ahead of his time. Possibly without realising it himself.

The same question, all day

Because if you look at where AI is heading, we are about to be confronted with precisely that kind of question continuously. Only not from a salesman, but from our own agents.

An agent that has prepared a quote. An agent that has assessed a supplier. An agent that has planned a project. An agent that has analysed a risk.

Our role shifts from doing to judging.

Approve or not? Continue or adjust? Yes or no?

And that sounds like a promotion. From doer to decider, who would not want that. But there is a catch, and it is the same one as with the salesman at the door.

A question without context is not a question. It is a guess with your name signed under it.

What has to be right before you can say yes

Before you can responsibly say yes or no, one thing has to be right: the context.

The agent needs to know how your organisation works. Which customers you have. What has been agreed. Which systems you use. Which choices were made in the past, and why.

Without context you get an answer. With context you get advice.

And across twenty of those questions a day, the difference starts to matter. A well-argued proposal you approve in thirty seconds. A proposal without reasoning costs you ten minutes of digging, or you click approve because it sounds reasonable and you have fourteen other things on.

That second thing is going to happen. Not because people are careless, but because thinking properly twenty times a day is not sustainable.

So the question is not whether your people are good decision-makers. The question is whether the proposals they receive are well enough argued to judge in thirty seconds.

What the coming years are about

Which is why I think the coming years will not be won by the organisations with the most expensive AI subscriptions.

They will be won by the organisations that make their knowledge, processes and data most available to the models they use.

That is not a technical task, or at most half of one. It is mainly the task of writing down what was never written down. Why that exception exists. What was agreed and with whom. Why we do it differently for that customer.

Exactly the things that sit in people’s heads, because everybody who needed them already knew.

With the right context, the question of whether he wanted to switch might have been very easy to answer.

We will never know.