Just Ask

You do not need to understand AI to use AI. You just need to ask.

The Thing We Keep Forgetting

I fell into my own trap last week. I needed a breakdown of billable hours allocated to a specific client line item. Without thinking, I pinged our analytics team. One of my directors looked at me with the patient bluntness of someone who has watched their boss forget the thing he tells everyone else to remember and said: “Did you ask Anton?”

Anton, for those of you reading this newsletter for the first time, is our AI agent/employee.

I had not.

I had Anton. We had built Anton. I gave a presentation to my team about Anton. I have written multiple newsletters about Anton. And yet, when I needed to figure out a way to use Anton, it didn’t occur to me first to ask the one thing that probably knew the answer better (or at least quicker) than anyone, which is Anton.

Instead, I defaulted to the old workflow and messaged a person.

This keeps happening. Not just to me. To everyone on my team. To the CEOs in Chicago who videotaped a basic AI workflow as if it were a magic trick. To my wife, who uses AI daily for things that are at times hilarious but mostly inspiring, and still, for certain categories of questions, reaches for a search engine or a phone call as if the past two years had not occurred.

Which means, when you think about it, the barrier to AI is not capability, access, literacy, sophistication, or understanding how language models work. The barrier is a reflex.

The entire adoption challenge of AI, the whole thing, reduced to its smallest unit, is this:

Remembering who to ask.

The Costco Test

A friend of mine recently went to Costco with his wife to buy a television, where he had the kind of disagreement that couples have in electronics departments, which is to say a disagreement conducted entirely through body language and strategic browsing. She went immediately to the three-hundred-dollar section. He went immediately to the OLEDs.

The salesperson began explaining the differences between the two models, which led to a divergence in enthusiasm, as it often does when specifications are recited to individuals. So he pulled out his phone, opened ChatGPT, and took a picture of the model numbers on the display. He then entered the specifications of the type of television that he has now and asked the AI the following question:

“We have this TV now. It lasted five years. We want something better. Help us decide between these two options. Ask me questions if you need to.”

Thirty seconds later, he and his wife were reading a side-by-side comparison that addressed the exact things each cared about (picture quality, lifespan, light reflection, whether the more expensive one was actually worth double the price or just marginally better).

But unlike the advancement of other technologies, my friend’s wife did not learn a new skill in that moment. What she saw was someone doing the oldest thing in the world. She saw someone ask a somewhat complicated question and get a somewhat easy-to-understand answer.

And that is what makes AI different than other technological breakthroughs. Unlike computers or the Internet, it is not a skill, but the simple ability to ask questions that will become the catalyst for how AI adoption actually spreads.

In other words, it’ll come through proximity. Through watching someone you trust do the thing you have not yet thought to do. Through the casual, undramatic moment where someone asks and the answer is useful, and you think: I could have done that.

You could have. You can. You just need to start.

The Two Kinds of Asking

But there are two kinds of asking.

The first is asking the machine. And it is remembering that, for the first time in technological history, when you don’t know how to use a piece of technology, you can actually ask it for help. For basic help, that is, but that is still a giant step compared to the way you’ve interacted with technologies before.

The second kind of asking involves asking a human.

Both are essential. Neither replaces the other.

And here is a great example of the difference between the two:

A prospective client called us recently. They had built a prototype with AI. They had a working interface, respectable enough to demonstrate. Whenever they got stuck, they asked the machine how to build it or fix it, and the machine answered, and the answer was good, and the prototype worked.

Then they called us.

Not because the prototype had failed. But because they understood that sometimes the answers are not enough. Sometimes what matters more is the person asking the questions.

A prototype is not a product. And that the gap between a working demo and a working business is filled with things the machine cannot anticipate: data compliance, accessibility standards, third-party failures, security vulnerabilities, the specific 3 am crisis that arrives because someone did something no one planned for.

They asked the machine first. Then they asked the expert. And when you put it that way, the prototype is not a deliverable. It becomes a question. And what I’ve come to realize is that the most articulate, useful kind of question a MarTech agency can be asked is:

“Here is what I am trying to build, here is how far I have gotten, and here is where I need help.”

This is the version of “just ask” that makes me hopeful about where this is all going. Because the client who brings me a prototype and says, “What would it take to turn this into production,” is not a competitor to my business. As a matter of fact, they are the best possible type of client because they have already externalized their vision in a way no client has ever been able to externalize their vision before. They have, in a sense, already asked the first question. And now they are asking the second harder one which just so happened to require a human on the other end who has been wrong enough times in this specific domain to know what wrong looks like before it arrives.

So, remembering to just ask is not only about the machine. It is about recovering the willingness to ask at all. The machine, then the expert.

To the School of Thought That Is Still Against Asking

One of the members of my YPO chapter told me about a form his son’s school sent home. A pledge for parents to sign, affirming that their ten-year-old would not use AI to complete homework assignments.

He stared at the form and felt something he could not quite name. It wasn’t anger. It was something closer to vertigo. The school was asking him to commit, on his child’s behalf, not to use the defining tool of the next several decades for the defining activity of childhood.

There are schools now, in several major cities, that are built entirely around AI-first learning. The students spend a few hours a day on curated, AI-assisted education. The rest of the day is projects: building things, making things, solving problems in the physical world. In a similar way, my clients approach me with their prototype, and the AI handles the information transfer. The humans handle the application.

He said he felt as if he simultaneously had a choice and did not have one. He could keep his son in the nice school that was teaching him to avoid the tool. Or he could consider a school that taught him to use it as the foundation for everything else.

I do not have children old enough for this decision yet. But I run a team, and the parallel is exact.

I can build an organization that treats AI as something to be cautious about, managed, and requires policies, pledges, and careful guardrails around who is allowed to ask and what they are allowed to ask for. Or I can build an organization that says, “Just ask.” Ask the question you are afraid makes you look stupid, because the question you do not ask is the one that costs you the most.

First, ask the machine. Then ask each other. Then ask me. Then ask the client.

I have chosen the second. Not because I am certain it is right, but because the alternative is an organization that has already decided to stop learning. And the one thing I am certain about is that this is not a moment to stop learning.

The Difference Between Asking and Knowing What to Ask For

When I asked Anton for the client hours breakdown, I got a useful answer because I knew which client, which line item, and which time period mattered. If I had asked for “a report,” I would have gotten a generic one. In other words, the specificity came from knowing the business.

But I’m not going to lie here. I’m not a CFO. The only reason I knew which questions to ask Anton in order to get the best answer was that I asked Anton which questions I should ask. As I mentioned before, this is one of the most amazing aspects of AI. And for most things that can help people do more things than they’ve ever been capable of doing before in their lives. But it also has severe limitations because, eventually, one hits the limit of knowing what to ask for.

This is the part that the “just ask” philosophy must not obscure: asking is the easy part. Knowing what to ask for is the hard part. And knowing what to ask for is not a skill you develop by using AI. It is a skill you develop by doing the work, by being in the room, by making mistakes that teach you which questions actually matter.

The tool accelerates everything that happens after the question is framed. It does not frame the question for you. Or rather, it can, but its framing will be generic where yours, if you have the experience, will be specific. And specificity is the difference between a search result and a solution.

I am not saying this to diminish AI’s ability to answer questions about how to use it effectively. As a matter of fact, I am saying it to protect it. Because if people believe the tool replaces the need for domain expertise, they will produce fast, confident, generic work. And the people who win will be the people whose questions are sharper, because their experience is deeper, because they have been doing the thing long enough to know which questions actually matter.

What I Have Learned in a Month

Four weeks ago, I wrote that AI had lost its wonder. Three weeks ago, I corrected myself: the wonder migrated, and I was maybe too saturated to notice. Two weeks ago, I wrote about the chat layer arriving to replace our interfaces, and this week I am naming the thing I have been circling since I started.

The single most important shift in my relationship with AI is not technical. It is philosophical. It is the decision, made slowly and imperfectly over the past year, to treat asking as the first step rather than the last resort. To prompt before I plan. To describe the outcome before I map the process. And then, when the machine has given me its eighty percent, I ask the next question, the human one. The one directed at the person who has the experience I lack, the scar tissue I have not yet earned, the judgment that comes from having been wrong in this specific domain enough times to recognize what wrong looks like from a distance.

Just ask. The machine first, then the expert. The prototype first, then the production team.

The revolution is not in the machine. It is in the habit of turning to it first, then to the experts.

It is a habit that has changed (and is still changing) my business, my team, and the way we think about products, interfaces, and what we owe the people who use or purchase the things we build.

Introducing: Ask Ashley

I have decided to take my own advice.

Starting next week, I am going to post a LinkedIn newsletter (their name, not mine) called Ask Ashley. It will be different from this one. Where When Simple Stops Working is exploratory, philosophy and argument, and the occasionally uncomfortable confession, Ask Ashley is going to be the practical side. The part where playing with the machine ends, and now you need to know more. The part where you have to begin asking a human being.

Every issue will start with a real question: something I asked, someone on my team asked, a client asked, or someone on LinkedIn asked.

I will show what happens when we ask the machine a question, and then how the human answered it afterward.

The goal of this newsletter will not be to lead to a course, a masterclass, or a promise to make you an AI expert. Or to even assume my answers are correct. My goal, instead, is for it to be a kind of running record of one company asking questions and sharing what comes back. Because I believe the single biggest barrier to AI adoption is not access, literacy, or cost. It is the simple, stubborn, deeply human reluctance to ask a question when you are not sure what the answer will be.

So I am going to do my best to answer these questions in public. Every week. And if the answers are sometimes wrong (they will be), or the prompts are sometimes clumsy, or the output requires more editing than I would like to admit, that is the point. Because that is the part where the human stays in the loop, and the work becomes, in some small and essential way, ours.

You can subscribe here if you want. Regardless, I will see you here next week.

And in the meantime: just ask.

Until next Wednesday,

Ashley Heron,
Managing Director, Comma Eight