Episode: Construction Conversations, Episode 65
Podcast Info: Construction Conversations Podcast
Chris Boyd on AI adoption: "The early adopters have done it"
Chris Boyd, Chief Product Officer (CPO) at Trunk Tools, joined episode 65 of the Construction Conversations podcast to talk about AI in construction. He covered where AI belongs next to the software builders already run, why so little of what one job learns reaches the next, and how to judge an AI tool by the work it does. The episode went live on October 20, 2026.
The short answer: AI in construction earns its place when it moves field work forward inside the tools teams already use, carries what one job learned into the next, and is measured one task at a time.
Picture a superintendent wrapping up a data center job. They know which detail caused rework, which trade needed extra hand holding and which sequence kept the schedule. On the next project, none of that shows up in the drawings. Chris made the same point on the show: "that's the type of stuff you don't catch on a drawing."
Where should AI sit in a contractor's software?
Contractors already keep drawings, specifications, RFIs (requests for information) and submittals in their systems of record. "We want to be the system of action that sits on top of a lot of those systems," Chris said. The job is to move field workflows forward using the data those systems already hold.
He also pushed back on how loosely the word "agent" gets used. Many tools sold as agents automate a single step. In his view, a working agent needs triggers that start it, workflows with several steps, evaluation of its output, monitoring, and verified project data to check its answers against.
Why doesn't construction learn from its own data?
General contractors (GCs) hold years of project records. "GCs have a lot of data they value, but they don't derive value from it," Chris said.
He traced part of that to what gets written down. "You've got a lot of plan, which is intent, but not a lot of actuals, which is what was actually built." Drawings and specs capture what the team meant to build. Far less captures what went in, so the lessons stay with the people who were there. Chris called this construction's closed-loop problem. One idea he raised: a running ledger of decisions that stays current as the job moves.
How should builders judge an AI tool?
"AI is not perfect. It's still a probabilistic system," Chris said. Many wrong answers, he explained, trace back to missing context, such as project files that have not synced yet.
His answer is to measure accuracy separately for each task, with test sets that include questions the system is expected to fail. The same discipline applies to the models underneath. Trunk Tools tests new models side by side against the ones already in use and swaps one in only when it does better. "It's about tracking the trajectory," he said, which means watching progress over months before judging any single release.
He closed on the people side. AI tools change daily habits, and he warned against teams dismissing what the AI flags without reviewing it first. Construction, he said, "has a change management problem." On adoption, he said: "The early adopters have done it. Now we're in the middle of that bell curve, and folks have got to get on board."
Frequently asked questions
Who is Chris Boyd?
Chris Boyd is Chief Product Officer at Trunk Tools, where he leads the product team.
Is Trunk Tools building hardware?
No. Chris said there are no plans to build hardware. He also raised privacy concerns about devices that are always listening.
Listen to the episode
Hear the full conversation on Construction Conversations, episode 65: constructionconversations.com/episodes/ep-065.html. Thanks to the Construction Conversations team for having Chris on the show.