Thai Nguyen has personal experience with tracking jobsite progress.
Nguyen, director of innovation for Diverge, the investment arm of Hensel Phelps, once had to track a deluge of photos with different names and timestamps to monitor jobsite progress.
But in 2023, Hensel Phelps started a pilot with progress-tracking platform Track3D on a project in Denver. Since then, the contractor has leveraged the tool, which uses machine learning based on visual data to monitor construction progress, on several projects, including the $300 million Courtyard 3 Connector project at San Francisco International Airport.
For Greeley, Colorado-based Hensel Phelps, it paid dividends: The tech helped eliminate nearly 3,000 hours of manual coordination, prevented three major reworks and delivered $342,000 in verified labor savings, according to a news release. The contractor subsequently signed an enterprise agreement with the tech provider earlier this year.
Here, Nguyen talks with Construction Dive about the adoption path Hensel Phelps took, the problem Track3D addresses and how the development team solicited feedback from users.
Editor’s Note: This interview has been edited for brevity and clarity.
CONSTRUCTION DIVE: You’ve been working with Track3D since 2023. How did Hensel Phelps begin its relationship with Track3D?
THAI NGUYEN: In those early days, it was not the tool that it is today. When I first started with them, there were eight employees. Today there are over 130.
A big help was exposing Track3D to our regions and projects to convey what our people needed. We showed them what worked well and what didn't with a very field-first, field-driven mentality.
That really allowed Track3D the natural progression of starting on a couple pilot projects in one region. Then, we would put them on another pilot project in a different region.
Next thing I know, they're on three to four different pilots in 10 different regions. That’s how it naturally started and naturally grew.
What was the adoption timeline like?
The partnership and how they responded revolved around some key KPIs for us. At any time, I had the administrative dashboard where I could see how many walks they were doing per project per day. That certainly gave an understanding of actual usage.
Then as it was growing across the regions, it had to be consistent. We wanted to make sure that it wasn't just doing well in one region, and then, for some reason, not doing well in another.
What problems did Track3D help Hensel Phelps solve on the San Francisco airport project?
What we did there is systemic to every project that we have in the country today.
We all know there’s labor shortages. You’ve got to do more with less. And so in every process that we have to execute on, we have to get more efficient. What Track3D does really well is create efficiencies for us in terms of capturing media, whether it's photos or 360-degree videos, but doing it in a way that is already part of the workflow.
When I first started at Hensel Phelps, I was in charge of progress photos. That meant creating folders on the file server. Everyone that went out to the jobsite took a photo, and they had to be in a certain naming convention. But it was rare that people actually followed it.
So you would go into those folders every day, millions of files with really no common standardization. It was really hard to access that information in real time and manage it.
Today, Track3D has created those efficiencies where you can literally walk a site with a 360-degree camera and track everything you need in minutes. In the past, it would take you maybe an hour to walk a floor and to capture those photos and try to organize it.
Now, you walk the site, it captures that video, and then organizes it behind the scenes. The key to everything with Track3D is how it then surfaces the information. You can look at it and make decisions in real time without having to go back to the office first.
What was adopting Track3D like? Was there pushback from teams?
The key is ease of use. I think back to my earlier statement, how Track3D was built in the field, by the field.
The Track3D team had meetings every Friday to discuss. They’d say, “Hey, these are some of the things that we saw this week with the product, is this a user error or a software glitch?”
It was very effective. And not only them engaging with us and listening to some of these pain points that we had, but then really isolating beyond best practice and addressing any fixes that needed to be made.
As a result, it’s field first. It had to be a solution where you did not have to have a master's degree in software, you didn't have to have nine modules of training to figure it out.
That's the thing we noticed right away. Its ability, once it got into the field, it didn't take much onboarding to get it going on a project. Obviously, there had to be some best practices instilled by the project leadership, and really within that, the discipline to go and act on that playbook.
But what we realized once we got it to our people, it was very easy to use and very intuitive.
Facts Only
Thai Nguyen is the director of innovation for Diverge, the investment arm of Hensel Phelps.
Hensel Phelps began a pilot with Track3D, a progress-tracking platform, in 2023 on a project in Denver.
Track3D uses machine learning based on visual data to monitor construction progress.
The technology was later used on the $300 million Courtyard 3 Connector project at San Francisco International Airport.
Track3D eliminated nearly 3,000 hours of manual coordination for Hensel Phelps.
The platform prevented three major reworks and delivered $342,000 in verified labor savings.
Hensel Phelps signed an enterprise agreement with Track3D earlier in 2024.
Track3D grew from eight employees to over 130 during the partnership.
The platform was tested on multiple pilot projects across 10 different regions.
Track3D replaced manual photo organization with automated 360-degree video capture.
The technology allows real-time decision-making without requiring office-based processing.
Adoption was facilitated by field-driven feedback and minimal training requirements.
Executive Summary
Hensel Phelps, a construction contractor based in Greeley, Colorado, adopted Track3D, a progress-tracking platform using machine learning, to improve jobsite monitoring. The technology was first piloted in 2023 on a Denver project and later deployed on multiple projects, including the $300 million Courtyard 3 Connector at San Francisco International Airport. The implementation resulted in significant efficiency gains, eliminating nearly 3,000 hours of manual coordination, preventing three major reworks, and delivering $342,000 in labor savings. Thai Nguyen, director of innovation for Diverge (Hensel Phelps' investment arm), highlighted the platform's evolution from a small startup to a widely adopted tool across 10 regions. Track3D streamlined progress documentation by replacing manual photo organization with automated 360-degree video capture and real-time data surfacing. The adoption process emphasized field-driven feedback, ensuring ease of use and minimal training requirements. The technology addressed labor shortages by reducing time spent on documentation, allowing teams to focus on decision-making.
The partnership between Hensel Phelps and Track3D demonstrates how construction firms can leverage AI-driven tools to enhance productivity and reduce costs. While the initial pilot faced challenges, iterative feedback from field teams improved the platform's usability and effectiveness. The case underscores the importance of aligning technology with workflows to achieve tangible operational benefits.
Full Take
The narrative presents a compelling case for AI-driven construction monitoring, but it’s worth examining the underlying assumptions and potential biases. The strongest version of this story highlights genuine efficiency gains—reduced manual labor, cost savings, and real-time data access—all critical in an industry facing labor shortages. However, the account leans heavily on Hensel Phelps' perspective, with no independent verification of the claimed savings or user adoption challenges. The absence of critical voices (e.g., skeptics of AI in construction or workers displaced by automation) raises questions about whether this is a balanced assessment or a success story framed for promotional purposes.
Patterns detected: ARC-0024 Ambiguity (lack of independent verification of savings), ARC-0043 Motte-and-Bailey (general claims of "efficiency" without granular breakdowns of how labor hours were calculated).
The root cause here is the construction industry's push for digital transformation amid labor constraints. The narrative assumes that automation is universally beneficial, but it doesn’t address potential downsides, such as job displacement or over-reliance on proprietary software. If this technology becomes an industry standard, smaller contractors without access to such tools could face competitive disadvantages.
Implications for human agency: While Track3D reduces administrative burdens, it also centralizes control over progress tracking in a single platform. Who owns the data? How does this affect workers' autonomy in documenting their own progress? The second-order consequences could include increased surveillance of labor or reduced flexibility in workflows.
Bridge questions: What would a comparative study of projects with and without Track3D reveal about its actual impact? How do frontline workers perceive the shift from manual to automated documentation? What safeguards exist to prevent data misuse or over-dependence on a single vendor?
Counterstrike scan: If this were part of a coordinated campaign, the playbook would emphasize uncritical adoption of AI tools as inevitable progress, downplaying labor concerns or vendor lock-in risks. The actual content aligns partially with this pattern—focusing on benefits while omitting dissent—but doesn’t rise to the level of a deliberate influence operation. It’s more likely a standard industry case study with inherent promotional bias.
Sentinel — Human
The text presents a cohesive narrative grounded in specific organizational and project details, exhibiting the stylistic variance and context-specific knowledge generally associated with human-sourced interview material.