Abstract representation of buildings and data.

Automating the Jobsite: Leveraging Artificial Intelligence for Risk Mitigation and Disputes

Contact: Marshall Harris

December 3, 2020

New software solutions to improve the development, organization and interpretation of jobsite documentation are becoming increasingly prevalent in the construction industry. These solutions often automate traditionally manual processes by utilizing artificial intelligence-based technologies to provide more comprehensive site documentation and provide deeper insights into existing project records. By making investments in these types of technologies to passively perform traditionally manual tasks, construction firms will gain insights that they can use to reduce project risk, improve performance, and ensure jobsite documentation is thorough and accurate for use in potential disputes that may occur.

Introduction

On a recent project, Ankura’s client, an established international builder, requested our assistance in quantifying the delays and associated increased costs on a large, residential condominium project. We found that the builder’s own project controls protocols had not been rigorously followed throughout the course of the project. Among other issues, the builder’s daily reports were incomplete, unreliable, or missing for large chunks of time; daily site photos had not been taken consistently, and the photos that had been taken were not labeled by date, floor or area; daily manpower tracking had been done manually, but was found to be incomplete or, at times, inconsistent with the site entry turnstile data; and “lessons learned” information from the firm’s prior, similar projects in the region had not been considered by the cost estimating team when the bid was being developed, which led to underestimating certain elements of the work.

Due to these issues with the client’s documentation and processes, supporting their delay claim was a significantly more arduous task than anticipated. While Ankura and the builder were still able to prepare a limited claim package, the builder’s management team agreed that more thorough documentation could have significantly bolstered the submitted request.

We also wondered: could our client have more effectively identified and managed its risks on the project contemporaneously if these key pieces of documentation had been readily available (and accurate)?  Our client agreed that if the information captured in these data sources had been readily available throughout the project, they may have been able to take action to avoid some of their own, self-inflicted issues – issues that the project’s developer surely capitalized on during claim negotiations.

All over the world, construction managers, general contractors, and subcontractors face a similar issue: they lack documentation and actionable information on their projects, and often this issue is the result of lapses by their own project teams. Frustratingly, this can prevent the same parties from taking timely action to avoid project impacts. And in some instances, claims for excusable delays and increased costs that would otherwise be valid are instead denied due to this internal failure to adequately document impacts contemporaneously, which results in a lack of substantiation when it is most desperately needed.

Fortunately, with the emergence of new software and hardware technology solutions to improve jobsite documentation and make use of existing, readily available data – on an automated basis utilizing artificial intelligence (“AI”) – many of these issues can be easily avoided. AI is a general term for tools that mimic human cognitive functions such as problem-solving, pattern recognition, and learning,[1] and AI itself is a broad category that encompasses a range of subset technologies, such as machine learning, natural language processing, and computer vision, among others.[2]

Despite the usefulness of these tools, a recent survey of 130 of the largest AEC firms in the United States identified that less than eight percent of respondents were utilizing AI technologies on more than a handful of pilot projects, and at least 46 percent of respondents were not utilizing AI technologies at all.[3]  With these technologies finally starting to mature, contractors and subcontractors should consider investing in solutions that automate traditionally manual processes and leverage AI to make actionable, real-time information available in a digestible format as a means to identify potential issues and/or impacts. The ability to do this quickly will, in turn, enable users to develop solutions that address the impacts in a timely manner, which ideally minimizes the need to submit a claims package at the end of the project.

Current technologies enable companies to do many things automatically, including recording progress on a jobsite over time, keeping track of the quantity and locations of personnel and equipment employed on the project, and making sense of the company’s historical cost and performance data. By making investments in these types of technologies to passively perform traditionally manual tasks, construction firms will gain insights that they can use to identify and reduce project risk, improve performance, and ensure jobsite documentation is thorough and accurate for use in supporting potential requests for equitable adjustment.

The Impact of Automated, Rich Visual Documentation

Today, the effortless, automated generation of comprehensive visual documentation enabled by new technology applications is a game-changer that has countless applications for enhancing real-time risk mitigation and improving dispute outcomes. These technologies often employ AI not only to capture the visual data but also to make the visual information easier to access, interpret, and take action on.

Frictionless, Structured Visual Data Capture

In one example, OpenSpace software enables construction personnel to effortlessly create comprehensive, 360-degree images of the jobsite and view the imagery in a web-based or mobile app-based platform that resembles Google’s “Streetview” interface. The imagery is captured passively via hardhat-mounted, 360-degree cameras worn by the people who are already on the site, such as the project manager, safety manager, or site superintendent, and the images themselves are tagged to locations corresponding to the project’s floorplans. Because the images are automatically captured continuously, viewers can then use the viewing interface to click through the rich images and “walk” the entire jobsite. Even better, viewers can use the interface to see how progress changed at a single location over time – even day-by-day, if needed. Issues identified while images are being captured can also be easily “tagged” to the location and the associated images, and the images even can be compared side-by-side to the corresponding location in the 3D design model if needed.

The availability of accurate and comprehensive jobsite photos has countless benefits for identifying and addressing risks contemporaneously. For example, during a jobsite meeting, it is simple to pull up nearly current imagery of any specific location on the project site to discuss and resolve a design issue that may impact the progress of the work. Photos captured via OpenSpace can also be easily employed when identifying issues that need resolution via Requests for Information or when raising a concern via an observation. Other functionality includes the ability to compare the images to the project’s 3D design model in real-time to identify discrepancies between the design and the installed work.

The 360-degree images can also easily be viewed over time to forensically document progress, identify the problem areas of a job, and compare progress against the planned schedule  when the necessary information or level of detail is not available in project schedules (or when project schedules are not even available). 360-degree photo platforms like OpenSpace, and others like OnSiteIQ, are starting to take this even further by using AI to identify the specific type and quantity of work installed at any given time, and by identifying when significant progress milestones are achieved, which can be used to confirm the accuracy of the schedules prepared on a project.

By relying on readily-available, 360-degree photos taken frequently during the project, it is also straightforward to determine when specific activities started and ended – or when an impact occurred – with a significantly higher degree of accuracy than would ordinarily be possible with the project schedules or other project records. The availability of such imagery may also provide the potential for greater specificity when quantifying the delays to a project’s critical path forensically, as monthly schedule updates often utilized to identify the critical path of a project are only a static snapshot, lacking insight into what occurred between the schedule updates that caused impacts.

In one example of this application, a drywall activity on the critical path of the schedule extended for fifteen days rather than the five days it was anticipated to occur, but the monthly schedule updates failed to provide any insight into what caused that extension or who was responsible. It would have been challenging and costly to identify the answers to those questions within traditional project documentation, but with the availability of comprehensive 360-degree photos, the exact area of work was able to be viewed on a day-by-day basis. This exercise provided clear evidence that nearly 95 percent of the drywall installation was completed within the planned five days, but an outstanding design question related to the in-wall mechanical, electrical and plumbing (“MEP”) rough-in activities in one specific work zone impacted the remaining five percent of the installation. The issue had even been contemporaneously tagged directly in the 360-degree images for that area.

In another example, during one project dispute in which it was alleged that there was defective installation of the roofing assembly, the regularly taken 360-degree status images helped identify exactly what product was installed, when it was installed, and how it was installed. This saved countless hours and the significant costs of investigations, and even avoided the need for destructive testing to occur, which would have required further costs and impacts to the building’s tenants.

Increased Value of Visual Data from Disparate Sources

AI technologies such as computer vision and machine learning are also permeating jobsite photo and video management and, in turn, allowing project management teams to take informed action based on visual information that typically sits on project hard drives disorganized and untouched. One company in this space, Smartvid.io, uses its AI engine, nicknamed “Vinnie,” to identify the indicators of project risk on a jobsite in the areas of safety, productivity, and quality. By aggregating photo and video content from disparate sources into one system, Smartvid.io enables project teams to easily access a dashboard that ranks safety risks and employs predictive analytics to reduce safety incident rates. The benefits of these insights can be substantial: on one project, it was identified that when Smartvid.io was utilized with the contractor’s own safety mobile application, recordable safety incidents were reduced by 28 percent, while lost time was cut by 35 percent.[4]

The visual search capabilities enabled by the “Vinnie” AI engine also allow for an expedited forensic review of the scopes that may be relevant in a dispute. The AI-based tagging system enables users to quickly search through images for specific scopes – for example, a search on the word “rebar” across all images and video on a project in which the rebar installation was alleged to cause delays – to help narrow down the pool of images that could be of use in analyzing an issue. Further, it is much easier to then determine pertinent information such as precisely when and where the work was done, how the materials were installed, and whether they were installed correctly. While it is possible to sift through months or years of general project photos or video files manually, relying on powerful machine learning-driven software to expedite such work reduces time and enhances efficiency and effectiveness.

Enabling Deeper Insights into Labor & Equipment

Construction innovators are envisioning a time in the near future when every person on a construction site is given a smart watch, smart hardhat, and augmented reality safety goggles that will work together to feed information in real-time to a project management command center. But even before such a seemingly science fiction world becomes a reality, by employing today’s Internet of things (“IoT”) hardware technologies such as wearables designed specifically for the jobsite, project teams can ensure they have access to real-time data that significantly enhance their understanding of what is happening on the project. Among other things, project management teams can use this information to improve safety, make timely decisions, or establish reliable support needed to substantiate or refute a claim.

Real-Time, Actionable IoT Data

Triax, one of the major companies in construction jobsite wearables, accomplishes these goals by providing clients with its personnel clips and equipment tags to enhance site visibility and security, improve worker safety, and enable productivity gains. The personnel clips and equipment tags work within a site-specific mesh network to contemporaneously identify the general location and activities of personnel and equipment. With this data readily available and easily accessible, it is possible to identify in real-time if an accident has occurred or which areas on a jobsite may be suffering from inefficiencies, such as the locations where there are more personnel than necessary (and in which those personnel are not accomplishing any more progress than anticipated).

By utilizing wearables and IoT tags, it is also easy to confirm – without ever leaving the field office – that the personnel or equipment specific to one scope are actually working in the areas where the project schedule anticipated they would be working on a given day. Furthermore, if equipment IoT tracking is being employed, it only takes a few moments to pull up the cloud-based dashboard to determine which pieces of equipment have not moved in the past few hours and are therefore being underutilized. The reasons for these deviations from the plan can be investigated, and project management teams can take action by reassigning personnel or equipment to a more productive task, by reducing staffing if necessary, or by removing equipment from the jobsite.

In practice, this approach has proven to be beneficial. A contractor utilizing Triax’s personnel clips discovered significant labor downtime due to a bottleneck of workers waiting for the construction hoist elevators. Triax hardware tags affixed to the hoists also enabled the contractor to gain insights into how the hoists were being used and, in turn, determine that the hoist was often being delayed at particular floors during the loading and unloading of heavy materials. The contractor was subsequently able to take action by dedicating two elevators to materials and two elevators to personnel, which helped to alleviate the labor bottlenecks.[5]

Improving Claims and Dispute Outcomes with IoT Data

Use of jobsite wearable technology could also have profound impacts for disputes and forensic analysis. The data generated by jobsite wearables can be used to automate the timesheet creation process,[6] which frees up project management personnel for other tasks. The accuracy of daily reports or timesheets can also be greatly improved by leveraging the data generated by wearable hardware systems. The data from wearables form an indisputable record of how many workers or items of equipment are on the project site and therefore improve upon the typical approach of developing daily reports manually by project personnel, which is often completed days after the work was done and can be prone to inconsistencies or lapses.

Moreover, automating these processes also ensures daily reports are not only regularly prepared but, most importantly, include accurate manpower counts and work locations, which can be key information for analyzing project impacts and supporting or refuting a claim. For the forensic analyst, the increased level of detail enables the ability to analyze the precise areas that are the subject of a claim or dispute, which is currently very difficult to piece together based on typically available project documentation.

In one example, Triax’s equipment tags were attached to swing stages that were used to perform façade repairs on a high-rise renovation project. The equipment tag data provided accurate, automated documentation of the utilization of the swing stages, which enabled real-time insights into the progress being made. Contemporaneously, this information enabled the project team to take actions such as deploying additional crew resources as needed to increase production.[7]  In the event of a dispute or for planning on future projects in which the progress on the façade scope was a key element to analyze, this documentation – which would be time-consuming to generate manually, if it were even tracked at all – could prove invaluable.

Automating Cost Processes and Insights

The lack of integration between various sources of information can be one of the most frustrating things on a jobsite. While this issue has improved over the past decade due to the emergence of software platforms that provide a single, cloud-based location for all of a project’s documents and data, there is still vast room for improvement – especially with cost estimation and accounting. Fortunately, currently available technologies are leveraging machine learning and natural language processing to provide automated insights into project costs. These solutions also enable estimating teams, project executives, and c-suite personnel to easily access information that may impact the cost of the work, which is still typically spread across various, fragmented documentation platforms.

Generating Cost Analytics with Machine Learning and Natural Language Processing

One solution establishing itself in this space is Briq, a construction financial forecasting and intelligence platform that uses machine learning and data analytics to provide contractors with insights on their project costs that help them to make decisions both on site and at the corporate level. By leveraging an abundance of readily available information – including both structured and unstructured data – from sources such as company accounting records and  public documents, Briq can identify trends in real-time that humans would typically either miss completely or identify only based on “gut feel.”

While projects are being built, contractors can rely on a software such as Briq to identify correlations in historical project cost data and even public records that no humans can do themselves without tremendous time and effort. The software delivers analytics and predictions into the data, which helps contractors to identify anomalies or validate the reasonableness of their own cost forecasting, among other applications. By providing real-time, early detection of cost anomalies, these insights can help a contractor identify an at-risk project early enough to take action that addresses the cost impact. The result is a powerful and much-needed bridge between a construction firm’s accounting and operations teams.

In a dispute situation in which a contractor submits a claim for additional costs, it would certainly put the contractor at ease to know it had utilized specialized software before the project started to confirm its cost forecasts were reasonable based its historical data, and that vast amounts of publicly-available records had been leveraged proactively to confirm any external variables were considered. Additionally, in such a dispute it could be instrumental to utilize AI software to help support the reasonableness – or proactively identify potential deficiencies – of a contractor’s bid or budgeted costs.

Looking Ahead

Market adoption of AI-based technologies is accelerating as the potential impact of these technologies is increasingly recognized, and it is estimated that by 2026 the global market for AI in construction will reach US $4.5 billion.[8] As technologies that leverage AI mature and become more widely accepted, the underlying algorithms will become more accurate and will therefore produce even better results. Exciting new applications of these tools for the construction industry are being developed right now, and some of these applications feature implementations that are even more “futuristic” than the examples described herein. Construction technology startups that leverage AI are being formed at a rapid pace, as evidenced in a recent report of construction technology trends, which identified that in the past five years the AI and advanced analytics segment had the highest proportion of new companies out of all the segments analyzed – and that trend was expected to continue.[9]

As exciting as future developments may be, the automation of traditionally manual processes using AI-based tools is already here, which makes this the right time for contractors to consider investments in these solutions. Applying these tools will help construction firms gain greater insights into their work, reduce project risk, improve performance, and ensure needed documentation is accurate and readily available in the event a dispute becomes unavoidable.

Many people recall that Amazon initially launched as an online bookstore – the first of its kind – and was among the first to embrace the potential of e-commerce as an approach to reaching customers. It spent the next 25 years investing in this concept to become the largest online retailer in the world, while thousands of brick-and-mortar retailers failed to adapt and innovate. AI-powered technologies present the same type of opportunity to leaders in the construction industry. Embracing these tools and investing in integrating this technology will provide construction companies  the ability to significantly improve their bottom line and expand their market share; but perhaps more importantly, those that fail to initiate adoption of AI-based technologies today may find themselves falling irreparably behind those that do.


Disclaimer: Reference in this article to any specific commercial product or service is for the information and convenience of the public and does not constitute endorsement, recommendation or favoring by Ankura. No warranty of any kind, implied or expressed, is given with respect to the products or services referenced herein.

[1] Rao, Sumana. “The Benefits of AI In Construction.” Constructible, January 18, 2019, www.constructible.trimble.com/construction-industry/the-benefits-of-ai-in-construction. Accessed November 3, 2020.

[2] Prieto, Bob. “Impacts of Artificial Intelligence on Management of Large Complex Projects.”  PM World Journal, June 2019, www.pmworldlibrary.net/wp-content/uploads/2019/06/pmwj82-Jun2019-Prieto-Impacts-of-Artificial-Intelligence-on-Management-of-Large-Complex-Projects.pdf. Accessed November 3, 2020.

[3] “BD+C Giants 300 Technology and Innovation Study 2019.”  Building Design + Construction, April 27, 2020, SGC Horizon LLC, www.bdcnetwork.com/accelerate-aec/exclusive-research-130-aec-giant-firms-reveal-their-top-technology-and-innovation. Accessed November 3, 2020.

[4] Gattie, Tim. “Strengthening safety culture at Suffolk with a new Observations app and a proactive process.”  Smartvid.io, December 18, 2019, www.smartvid.io/ai-in-construction-blog/strengthening-the-safety-culture-at-suffolk-construction. Accessed November 3, 2020.

[5] “AECOM Tishman Raises the Bar for Jobsite Safety and Productivity.”  Triax, www.triaxtec.com/wp-content/uploads/2020/04/Aecom-CaseStudy-1.pdf. Accessed November 3, 2020.

[6] “Save Up to 214 Man Hours Each Month with Automated Time and Attendance.”  Triax, www.triaxtec.com/wp-content/uploads/2020/04/AutomatedTimesheets-UseCases.pdf. Accessed November 3, 2020.

[7] “AECOM Tishman Raises the Bar for Jobsite Safety and Productivity.”  Triax, www.triaxtec.com/wp-content/uploads/2020/04/Aecom-CaseStudy-1.pdf. Accessed November 3, 2020.

[8] “Artificial Intelligence (AI) in Construction Market to Reach USD 4.51 Billion By 2026 | Reports And Data.” GlobalNewswire, July 23, 2019, www.globenewswire.com/news-release/2019/07/23/1886563/0/en/ Artificial-Intelligence-AI-in-Construction-Market-to-Reach-USD-4-51-Billion-By-2026-Reports-And-Data.html. Accessed November 3, 2020.

[9] Bartlett, Katy, et al. “Rise of the Platform Era: The Next Chapter in Construction Technology.”  McKinsey & Company, October 30, 2020, www.mckinsey.com/industries/private-equity-and-principal-investors/our-insights/rise-of-the-platform-era-the-next-chapter-in-construction-technology. Accessed November 3, 2020.