Building Your Own AI Tools? How to Navigate Critical Decisions with Confidence

As someone who has the privilege of working in customer success for a construction software company, I’ve learned that the most valuable thing my team and I can offer customers right now might come as a surprise. In fact, it doesn't always have to do with our product or even technology in general.

So what is this big secret? It's empathy.

Over the years, I've developed close relationships with contractors who are building the infrastructure that shapes communities: roads, bridges, utilities and so much more. These are the projects people rely on every day, and while it's no doubt exciting that AI is opening doors nobody imagined five years ago, it's also a lot to carry.

Because I spend my days talking with these construction professionals, from companies big and small, I tend to see both sides of the AI moment at once. On one side is the drumbeat everyone knows by now: We need more AI, faster. On the other side are those same people, often in the same conversation, quietly wrestling with what the excitement actually asks of them. Do I walk away from the tools I know and trust, build something myself and hope it holds up?

There is real enthusiasm here, and there is genuine unease. Lately, there is also growing fatigue as the wider world discovers that AI has not solved everything it promised and has sometimes cost more than anyone expected.

What I hear from contractors, though, is rarely fear. It is uncertainty.

They are not lying awake worrying about software features. They are losing sleep wondering whether they are making the right call. They are spinning plates, running crews, managing margins and weighing which technology to adopt and when, all while trying to guide their businesses into a future none of us can fully predict. The pressure surrounding AI stems from the fact that no one can tell them, with any real confidence, what they should do next.

Vibe Coding Your Tools Is a Bigger Gamble Than It Looks

That tension shows up most clearly at the decision points. As costs rise across the board and contracts come up for renewal, a contractor naturally starts weighing the alternatives.

Increasingly, one of those alternatives is building something in-house. Our team often hears some version of, “We’re going to vibe-code our own timecard system,” using the shorthand that has emerged for building software with the help of AI.

That decision often comes from an executive rather than a technologist. The reasoning is perfectly understandable. Executives want to cut costs, improve efficiency and gain better visibility into the data. But once the decision is made, the real work may fall to someone else with a deceptively simple directive: Build us something that does the job.

Whenever I am in one of those conversations, my instinct is not to argue against building. Sometimes building really is the right answer. What I try to do instead is widen the lens to the things that are easy to miss in the excitement, starting with data.

What you give a system is what you get back from it. Mature tools carry years of accumulated, hard-won knowledge about the specific ways this work goes sideways. When you start from a blank page, you inevitably start without all of that, and you cannot plan around the problems you have never had to see. This data question deserves to sit at the center of the decision, because the quality of what a system is built on largely determines whether you can trust what it gives back.

The second issue is the gap between building something and running it.

You might well develop a working iPad-to-cloud timecard app over the course of a weekend and, for a moment, it can look like the problem is solved. But rolling that out to every field crew, supporting it when something breaks at 6 a.m., maintaining it as your needs change and building airtight security around it is a different mountain entirely.

What looks like a one- or two-month project has a way of becoming a one- or two-year headache before it can truly replace the system you already had in place.

You Don't Know What You Don't Know

Then there’s the question of reliability, which matters most in our line of work. The software a contractor runs is directly tied to whether the business survives. AI systems can be wrong, and they can be wrong in ways that are difficult to see coming.

That’s why it’s fair to ask whether you are prepared to put the future of your company in the hands of a solution you cannot open up and fully understand. The goal is to know exactly what you are trusting, and why.

The stakes are even higher for anything safety-related. This is dangerous work, and a tool built to track compliance or protect the people on a crew carries real weight. The higher the stakes (safety, compliance, cost, project risk, etc.), the more scrutiny a tool warrants, and the more important it is that reliability, security, governance and human oversight are in place. Handing that responsibility to a system nobody can see inside or control is a far heavier decision than it appears in the moment of the pitch.

The point is to surface the practical realities that AI excitement can push out of view. And that, more than anything, is where a good partner earns a place at the table.

What directly helps a contractor in this moment is a partner who adds real value and who also sits on the same side of the table, someone who says, "I get why this feels hard. Let's figure it out together." The most useful thing I can offer is not a sales pitch or a warning. It is helping someone see the whole of the decision in front of them, so that whatever they choose about which AI platform to trust and adopt, they choose it clearly.

The future is uncertain for all of us. Facing it alongside someone who genuinely cares? That changes everything.