Most leaders come to AI with a shopping list. Which model, which platform, which tool everyone else is using. It is the wrong question, and it is costing them the answer. The leaders who get somewhere are not asking what to buy. They are asking what is actually broken.
I learned to say this plainly only recently.
When people ask what I do now, I have stopped reaching for the consultant language. No "AI adoption strategy." No "operational streamlining." I tell them the truth, which is shorter and more accurate.
I solve problems.
Here is the one that taught me to say it that way.
The Problem Nobody Had Named
A commercial construction firm brought me in. Good company, real scale, the kind of operation that runs on dozens of active builds at once.
Their project managers live in the field. They are on job sites, in trucks, on their feet all day. They do not sit at computers, and most of them would tell you straight out that they are not technical and have no interest in becoming technical. That is not a flaw. It is the job. You do not want the person running a build staring at a screen.
But the company's project management software needed those field managers to log updates. Stage of the build, what got done, what is behind. The customer success team depended on that information to do their own work, because customer success is the team that calls the client with the right update at the right moment. Where is my project. What happens next. When.
The field managers were not logging the updates.
They weren't doing it out of defiance. The software was built for someone who sits at a desk, and they do not sit at a desk. So the data never made it into the system, and the customer success team was flying blind. They were calling clients with stale information or chasing field managers for status that should have been a click away. Two teams, quietly frustrated with each other, both convinced the other one was the problem.
And here is the part that matters most.
Nobody had named it as a problem. It was friction everyone had learned to live with. It did not show up on a strategy deck. It was not a line item anyone was trying to fix. It was just the way things were, the cost of doing business, the thing people complained about on Fridays and forgot about by Monday.
I did not walk in knowing any of this. I found it the only way you can find it, which is by sitting down with the team leads and the executive leadership team and understanding how the work actually moves through the company.
I Wasn't There to Sell a Tool
Somewhere in that conversation, it clicked for me.
I was not in the room to recommend a piece of software. I was in the room to understand an operation. And once I understood it, the shape of the solution was obvious in a way it could never have been from the outside. You cannot buy your way to a fix for a problem you have not seen.
That is when I realized how I would describe my work from then on.
I solve problems. AI is the material I build the solution out of. But the solving starts with understanding, and the understanding starts with the kind of question most leaders skip right past on their way to the buying decision.
What If the Field Managers Never Opened the Software
So I asked a different question. Not "how do we get the field managers to use the software." That question had already failed for years. The better question was: what if they never had to open it at all?
What if, at the end of each day, an AI agent simply texted every field manager?
A short back and forth. Two minutes, on the phone already in their pocket. What did you wrap up today. What stage is the build at. Anything blocking you. The kind of conversation a field manager would have with a foreman without thinking twice. Then the agent takes that exchange and writes it straight into the project management software, structured and clean, no keyboard involved.
The customer success team wakes up to current status on every active project. No chasing. No guessing. No stale calls to clients.
The field team never touches a screen. The data shows up anyway.
Now, the honest part, because I will not tell you a story with an ending it has not earned yet. That agent is an idea on the table, not a system that is running. It can be built. We are working through whether it is the right build for them and what it costs to do well. And it is not in motion, because we are doing the part that comes first: training their executive team and writing their AI policy and governance before a single workflow gets automated.
That sequence is deliberate. You do not bolt an autonomous agent onto a company that has not decided what its rules are. The build waits. The foundation goes first. Always.
But the thinking is the point. In one conversation, a problem that had quietly taxed two teams for years turned into a build that a company of almost any size could reach, with the right thinking partner in the room.
Tool-First Is Backwards
Here is the trap, and most leaders are standing in it right now.
They start with the tool. They read about an AI platform, they see a competitor announce something, they feel the pressure, and they ask their team: what should we buy. The whole conversation organizes itself around the product before anyone has named a problem worth solving.
No tool purchase would have fixed the construction firm. Not because the tools are not good. Because the problem was invisible from the org chart. It lived in the gap between how the field works and how the software assumed the field works, and you cannot shop for a solution to a problem you have not put into words.
Tool-first thinking gives you a drawer full of subscriptions and a team that still does the work the hard way. Problem-first thinking gives you a build that removes friction someone actually feels.
The order is not a detail. It is the entire difference between organizations that see a return on AI and organizations that have a lot of logins and not much to show for them.
Where the Work Actually Lives
Go one layer down, because this is the thing I want you to sit with.
The field managers were never the problem. The customer success team was never the problem. The software was asking the field managers to become people they are not, and they reasonably declined. The friction was the distance between where the work actually lives and where the system expected it to live.
The field managers live on a text thread and a phone call. That is their native environment. The old answer was to drag them into the software's environment, and it failed for years because you cannot train a person out of the way their job actually works.
AI let us flip it. Meet them where they already are. Let the agent do the translation between the human environment and the system environment, so nobody has to become someone they are not.
That is not a story about construction software. That is a pattern, and once you see it you start seeing it everywhere in your own organization. The places where you have been trying to change the people because it seemed easier than changing the system. AI quietly makes changing the system the easier path.
You Don't Have to Build It Yourself
There is a fair objection sitting in the back of your mind right now, especially if you are leading a mid-market company without a deep technical bench. This sounds great, but I do not have engineers to build agents.
You do not need to.
When a client does not have the internal infrastructure or the in-house knowledge to build, myself and my team of thirty engineers do the building for them. The strategy and the problem-solving is the part that requires understanding your business. The build is the part that requires people who do this all day, and that part is available to you whether or not you ever hire a single technical person.
The unit of value here is not a platform overhaul. It is the micro-build. The texting agent for the field team. The small, specific solution to the specific friction, built well and governed properly, repeated across teams and departments until the easy workflows you never thought were possible become the way things run.
These are the conversations I look forward to most. Not the grand strategy decks. The rooms where an executive team starts naming what is actually broken, and we start sketching the small things that fix it.
Start With What's Broken
So here is what I would have you do this week, and it does not cost anything.
Stop asking what AI you should buy. Walk the floor instead, literally or figuratively. Sit with the people doing the work. Ask them what is hard. Listen for the friction everyone has stopped noticing, the thing people complain about and then shrug off, the gap between two teams that each think the other is the problem.
Write those down. Those are your candidates.
And when you hit one and cannot tell whether AI can help, that uncertainty is the signal. Not to buy something. To bring in someone who understands what AI can do, what it cannot, and how to build it properly before you commit to anything.
Because the tool was never the hard part. The tool is a short decision once you know the problem.
Knowing the problem is the whole job.
