Someone is pitching you a custom AI solution right now. Or they will be soon. The pitch sounds compelling: your data, your workflow, your competitive edge, built to spec.

Before you write the check, there’s a question worth asking. What problem, exactly, are you solving?

Start with the problem, not the pitch

The AI conversation in the built industry has gotten backwards. Firms are buying tools looking for problems instead of finding problems and then choosing the right tool.

Custom-built AI solutions are genuinely powerful when they’re solving a specific, well-defined pain point, something your existing software doesn’t handle, something that directly affects project margins or owner satisfaction.

They’re expensive distractions when they’re built because someone thought it would be cool to have a custom bot.

The question isn’t build vs. buy. The question is: what specific outcome are you trying to produce, and what’s the most practical way to produce it?

The real cost of “building”

Low-code AI tools have made it genuinely possible to build functional software in hours without an engineering team. That’s real. It changes the calculation.

But there’s a hidden cost that never shows up in the initial pitch: maintenance. Every custom-built tool requires someone to maintain it. Someone to update it when the underlying AI model changes. Someone to fix it when the data structure changes. Someone to train new users.

If you don’t have someone who can own that, you don’t have a tool. You have a liability.

“Adoption is your signal” is the most underrated test. A tool that goes unused is worse than no tool at all. It consumed resources, created expectations, and delivered nothing. Give people low-stakes tools first. Watch what actually gets used. Build from there.

What this means for owners

You’re on the receiving end of your partners’ technology choices. When a general contractor builds a custom cost-tracking tool that only their team understands, your access to project data depends entirely on whether they share reports. You don’t have visibility. You have a report.

That’s a different thing.

The AI investment decisions your design and construction partners are making right now will shape how much transparency you have on your projects for the next several years. Whether they’re investing in tools that give you real-time data access or tools that create more sophisticated ways of delivering information after the fact.

Ask them. What AI tools are you deploying on owner-facing workflows? How does it improve my access to project data? What does my project information look like in real time, vs. what gets packaged into a report?

The firms investing in AI that benefits owners, not just themselves, are signaling something about how they think about the relationship.

The practical framework

If you’re evaluating AI investments for your own organization, or benchmarking your partners’ approaches, here’s a simple filter:

Does this tool solve a problem that’s currently costing real money or real time? Can you measure the outcome? Will the people who need to use it actually use it, in their existing workflow, without requiring significant behavior change?

If the answers are yes, the build vs. buy question is secondary. Pick whatever gets you to a working solution fastest.

If the answers are unclear, slow down. The tool isn’t the bottleneck. The problem definition is.

Speed matters. Waiting for the perfect solution while your competitors ship imperfect-but-working ones is a losing strategy. But moving fast in the wrong direction just gets you lost faster.

Start with the problem. Work backwards to the tool. Don’t fall in love with the technology for its own sake.