No is not an answer. At most: not like that, but like this

And as the name appeared on my display I knew exactly what it was about.

The timing was not great, and not because I had no time. I was not on another call, and in principle that is the only moment I have no time.

The feeling came because I already knew I had no answer to the question I thought was coming. If I had one, I would have had all the time in the world.

So shouldn’t you just say that? And “manage the expectation”?

Yes, of course you should. But pleasant it is not. Partly because sometimes you know that you, or somebody else, already managed that expectation to death for nothing.

Once I picked up, I turned out to be right

A question in search of an answer. A solution. A solution to a problem several of my colleagues had already tried to solve.

The same goes for them. If they had an answer, they would have given it gladly. That is what we do it for; doing somebody a service. That is rather the point of being a service provider, no?

And no is not an answer at all. Not in our trade. Not in a world where robots drive around on Mars. (The distance to Mars is a minimum of 60 million and a maximum of 400 million kilometres, for those keeping score.)

At most it is: no, not like that, but like this.

And so it went.

“Very annoying. And yes, it takes a long time for something that looks so simple. But there is always a solution. They do drive a robot around on Mars, after all.”

One of my one-liners. Like the other one I have been using for fifteen years and which some people now consider a platitude: we need to know what the real problem is, and then the answer is usually very simple.

And to know that, you have to ask the right questions

That is the whole thing, and it is less obvious than it sounds.

It is about context. Every question hands you context back. And with that context you arrive at an answer.

Have you restarted your computer? Is there an error code? Are you the only one in your area with this? Since when? Which version?

Nobody enjoys those questions. Not the customer, who wants a solution and not an interrogation. And not the engineer, because every question is another round of waiting. So we skip them, we guess, and three days later it turns out to have been something else entirely.

Our trade loses more time that way than on all the complicated outages combined. Not because we do not know, but because we want to arrive at the answer too quickly.

What I did about it

I built an agent that helps you ask questions.

Give it a problem, and like a four-year-old it wants to know everything. Answer those questions and you get more questions. Right up until the problem is solved. The more context it has and the more it learns, the better the odds it solves it for you.

That is not a clever AI application. That is the patience I do not always have myself, outsourced to something that does not mind. An agent does not get bored at the seventh question. It does not find it awkward to ask one more time since when this has been happening.

And that is exactly why it works. Not because the model is smarter than my colleagues. Because it has no urge to start answering yet.

And of course there are humans in the loop. Of course you should just call if you have a question. But with the technology we already have, our daily work is going to look genuinely different.

So; got a problem?

Talk to my agent. ✌🏻