The State Of AI in Construction

Adam Krob, Director of Information Technology, Field Audit, and Process Improvement, Boh Bros. Construction

Through this article, Adam Krob, Director of Information Technology at Boh Bros. Construction, outlines five categories of AI adoption in construction—from tools delivering clear value to those still searching for impact. He shares practical examples, including AI that speeds up takeoff processes, progress tracking with mixed results, and meeting transcripts that boost team efficiency. Adam also warns against “AI washing” and stresses the need for clear KPIs and strong IT governance to unlock AI’s actual value in the industry.

The construction industry has been accused of being very late to derive value from technology solutions. Not without reason. Deloitte’s often-referenced report on the impact of technology on productivity by industry ranks construction as next to last, only beaten to the bottom by non-farm agricultural production – hunting. AI represents a leapfrog opportunity for construction to catch up to and potentially surpass other industries while, hopefully, maintaining its lead on hunting.

Five categories of AI solutions

Construction-oriented AI solutions vary widely, from those demonstrating value to those with no clear path to value.

There are five categories of solutions.

1. Delivering value - The first category includes mature solutions that deliver tangible value. These solutions have well-developed KPIs and documented successes.

2. Value is beginning to emerge - The second category includes solutions that are still trying to demonstrate their value. There are relatively well-defined KPIs, and successes have been mixed.

3. Still quantifying the value - The third category is the largest category of AI solutions with suggestions of KPIs (for example, resource reallocation), but they have not been demonstrated.

4. The value is not yet clear - The fourth category contains AI solutions where organizations are experimenting to explore the function and potential return on investment.

5. No clear path to value - The final category of AI solutions in construction is those where the solution provider has added AI to a product, the addition of which provides little tangible value. The following case studies illustrate solutions in three of these categories.

“While AI solutions are beginning to demonstrate value, challenges and risks remain. Companies must start their AI journey with well-defined IT security and information governance policies to act as guardrails”

Case Study I – AI-driven takeoff delivers value.

My company evaluated five AI solutions against our current processes about a year ago. We chose three with the highest potential value and the best complement to current processes. One of the solutions with the highest combined scores was a takeoff product that used AI to automate shape outlines, identification and counting of drawing objects such as doors or electrical boxes and provide initial square footage and volume calculations. After a short pilot, the team determined that AI integration into the takeoff process reduced the time to completion, allowing faster responses to project changes and more frequent retake-offs. Two other positive outcomes came from the pilot: the reduction in license costs from consolidating to a single, more powerful takeoff solution and the reduction in training costs and time to proficiency to learn the takeoff process.

Case Study II – AI-enabled progress tracking shows potential

For several years, my company has used 360 cameras and specialized software to record walks the project team uses to review and document progress. Over the past three years, solutions have added AI-based progress notifications. Over the past 18 months, the project team at one of our vertical construction projects piloted a competitor to our solution that incorporates this progress notification. The team determined that the new solution was very effective in notifying the project team of subcontractor progress for some scopes of work, such as drywall, but not others, like mechanical. In addition, the solution cost was greater than the hourly cost of a project engineer or junior superintendent walking the site at the same frequency. As the costs decline and the models improve, solutions like the one piloted will deliver value, but that value is still emerging.

Case study three – AI-created meeting transcription value still requires measurement

Our company's AI innovator community of practice has fully embraced AI-created transcripts for meetings with internal teams, subcontractors, owners, and architects/designers. The users of the solution all highlight the value of the solution, as letting the AI take notes 1) reduces the time to process meeting notes – as some meetings required more than an hour of work to document, which has been reduced to a few minutes, 2) frees the note taker to participate in the meeting actively, 3) allows team members not to attend if they were only participating as an “informed” stakeholder, and 4) builds greater accountability by ensuring that action items are documented, communicated, and available for planning  the next session. The team has not settled on a specific, measurable outcome. Still, it continues to document how this time is utilized for other purposes, such as a former notetaker conducting an additional safety inspection or quality check.

Risks remain

While AI solutions are beginning to demonstrate value, challenges and risks remain. Companies must start their AI journey with well-defined IT security and information governance policies to act as guardrails. In addition, many software vendors are adding AI to their offerings without a strong connection to the core value of their products. Distinguishing real AI value from “AI washing” takes time, effort, and serious consideration. The most effective way to judge is whether AI will deliver defined and demonstrated value.