Greg McClelland, co-founder and partner at Onetrace, discusses the best use of Artificial Intelligence in digital construction
Like any other tool, AI is sometimes the right fit for the job at hand, and sometimes it’s like using a hammer to screw in a lightbulb. With AI increasingly built into the digital infrastructure we use every day, it’s important to know when to pull it out of the toolbox and when to leave it alone.
For main contractors and tier-one supply chains, the pitch for AI is compelling: faster compliance checks, smarter defect detection, and leaner project reporting. While many are grappling with the complex task of optimising AI, organisations of this size can at least typically rely on having the time and resources to invest in specialists who can tailor the technology to their teams and goals.
Likewise, AI offers obvious advantages for subcontractors. The efficiency benefits could really boost businesses with thin margins and tight schedules that spend a lot of time on data-based tasks such as quality assurance, documentation, and environmental compliance. However, given the novelty of the tech and its rapid evolution, it can be confusing to know exactly when to use it.
Understanding the general value and drawbacks of this tech, along with the practical realities of adoption, is essential to unlock its potential while avoiding pitfalls.
Savings and productivity
The most frequently touted perks of AI for subcontractors are lower costs and higher productivity. Where traditional cost-management methods can struggle to keep pace with fluctuating prices, AI tools can deliver speed and precision. Using historical and contemporary data, these platforms can create accurate estimates and apply real-time monitoring to make swift adjustments, helping subcontractors stick to a budget despite rising labour, material, and compliance costs.
In areas such as project planning and scheduling, materials procurement, and safety, AI-enabled tools can deliver productivity gains. Examples include using AI tools for materials selection and quoting, or using AI to support logistics, scheduling, and resource allocation to ensure materials and machinery arrive on-site precisely when needed.
Verifying the work that actually gets done
A practical example of the opportunity AI presents for subcontractors is in more comprehensive install verification. Right now, only a fraction of completed works are physically inspected on-site because resource constraints mean quality checks are sampled rather than comprehensive. A supervisor can only be in so many places, and walking every metre of cabling, every fixing, or every joint on a large project is rarely feasible within the time and headcount available.
AI-powered image recognition and sensor data are beginning to change that equation. Instead of a human inspector sampling a percentage of completed installations, AI systems can analyse the full scope of work and automatically flag deviations. In practise, an on-site operative provides a timestamped record of the work with photographic evidence, and the AI produces a record of whether that install was checked and the outcome – building a quality assurance history directly into the installation record. For subcontractors, this offers the prospect of 100% install verification without 100% of the labour cost, a genuine step forward for quality assurance and liability management alike.
Cleaner data with less admin
Construction generates a mountain of paperwork, particularly around health and safety or environmental compliance. As BREEAM notes, data quality and assurance are core to environmental management and compliance, and AI is proving useful in making that process more efficient and accurate.
Capturing data on a construction site can be slow, and manual entry into reporting platforms is rarely as accurate as it should be. AI-enabled apps can help tackle this challenge: a site worker can photograph a document on a smartphone or tablet, and an AI-powered optical character recognition model can scan the image, extract data automatically, and prefill most of the information in whatever format is required.
A verification stage is particularly important, and it’s what separates fit-for-purpose AI tools from simpler digitisation efforts. Ultimately, the dashboards and reports a business relies on are only as good as the data feeding them, and pairing automated extraction with a human check keeps accuracy high without sacrificing time. For a subcontractor juggling multiple sites and constant paperwork, that combination translates into real hours saved and fewer errors creeping into compliance records.
Real-time updates
AI is also reshaping environmental oversight. Combined with IoT sensors, it allows continuous, real-time monitoring of critical site metrics such as air quality, noise levels, and water usage. Rather than waiting for a scheduled check or a complaint, teams get constant surveillance that flags any deviation from the norm the moment it happens, allowing an immediate response before a minor issue becomes a compliance breach.
Dubai’s Burj Khalifa shows how AI powers real-time updates: an AI-driven maintenance system monitors the building’s 57 elevators, eight escalators, and other machinery across 163 floors, detecting even the slightest signs of machine failure. Similar systems can give teams advanced maintenance warnings by proactively predicting when onsite equipment will need repairs or replacement.
Recognising the risks
Yet the risks are equally real, and in an industry where mistakes can be structural rather than cosmetic, they deserve just as much attention as the upside.
One of the most fundamental risks is that AI systems can hallucinate, creating confident, plausible, and entirely wrong outputs. A model analysing an image for a fire-stopping installation, or a tool summarising a load calculation, can produce an answer that reads as authoritative while being subtly, or completely, incorrect. Unlike a human error, which often comes with some hesitation or a flagged assumption, an AI’s mistake arrives with the same fluency and certainty as its correct answers.
In a regulated environment where a missed safety check can have fatal consequences, that matters enormously. Blind trust in algorithmic outputs risks putting efficiency ahead of due diligence, a trade-off subcontractors can least afford given that liability often lands on them when something goes wrong on site.
This further underlines why any deployment must preserve human sign-off as the final authority. AI can sample more, flag more, and process more than any team of inspectors could manage manually, but the decision that a piece of work is compliant, safe, and complete must remain a human judgement. That means treating AI output as a recommendation or a flag for review, not an automatic pass, and making sure the people signing off understand the tool well enough to know when to be sceptical.
Knowing where the data goes
Alongside risks to build quality and reporting, the data itself needs protecting. AI platforms require a lot of information to get the most out of them, and much of what a subcontractor handles is sensitive.
Workers onsite could accidentally cause data breaches, for example, by using AI to process a photo of a site document, inspection record, or sensor reading that contains more than the intended data point, such as location information, other parties’ commercial details, or personal information about colleagues. Without proper safeguards for storage, access, and retention, a subcontractor could accidentally become responsible for a serious breach.
Many businesses also want to move information between platforms to build bespoke dashboards or meet different reporting demands. AI can improve data portability, letting teams import third-party project data into their own records and analytics. This simplification is often welcome, but it means subcontractors need to think carefully about what happens to their data once it leaves their hands and while an AI tool processes it.
Keeping an eye on costs
While one advantage of AI is its potential to save subcontractors money, it’s worth noting that this technology comes with its own costs that need to be factored in. Recently, AI platforms have been moving away from seat pricing, where each person pays a fixed subscription regardless of how much they use the tech, toward usage-based or token billing systems. This shift will make AI more expensive for any business that uses it heavily.
As AI adoption is in its infancy, the way platforms choose to bill is still in a state of flux, and it can’t be assumed that an AI budget today will get you the same amount of computing power tomorrow. This means the commercial cost-benefit analysis must be reviewed regularly to ensure it remains in your favour.
Team-ready technology
To use the tool analogy again, there’s no point choosing equipment that your team doesn’t have the skills to use, or for which the required training is prohibitive on cost, time, or expertise.
That’s not an argument for avoiding upskilling altogether, but for considering what is reasonable and practical to build into your immediate processes while keeping an eye on what future projects may require.
Regular horizon scanning is important, as AI adoption will only become more integrated into the industry over time. A key driver will be tier-one contractors who need aligned data to feed their own AI-powered compliance and reporting systems, which are, in turn, fed by subcontractors doing the installations and on-the-ground work.
The fact that the number of UK contech businesses has grown by more than 200% over the last decade highlights how the market is adapting to the need for enhanced digital support to deliver modern developments
Overcoming adoption challenges
Despite growing AI awareness and contech options, adoption rates struggle to keep pace with the scale of the challenge, and it’s not uncommon for subcontractors to still run operations on paper timesheets and spreadsheets.
The reasons behind this slow take-up are multifaceted. Research has found that two-thirds of adults in construction couldn’t complete all twenty tasks that industry and government agree are essential for work, things like using Teams or Trello, compared to around 50% across UK industries generally. That makes sense for a workforce that rarely sits at a desk juggling applications all day, but it is a reality that needs addressing to ensure the sector is ready for tomorrow’s tech demands.
Further research from Cornell University identified seven barriers to contech adoption that highlight many of the challenges subcontractors face. Chief among these is a persistent skills gap, as many firms simply haven’t invested in the systematic training and education needed to build competency across their workforce, leaving even willing adopters ill-equipped to use the technology effectively.
This is compounded by a deeper cultural resistance to change, as individuals and organisations accustomed to traditional workflows are often reluctant to embrace new methodologies. This can be a particular issue for smaller firms where the friction of change often outweighs the theoretical benefit of switching.
For data-based systems, securing buy-in from teams on the ground is essential, since they gather the data the AI depends on. If they don’t know how to record information accurately, or have no incentive to bother, the system’s value collapses, no matter how good the underlying model is.
Economic factors reinforce this inertia. Many in the sector remain unconvinced of contech’s tangible business value, unable to see a clear line from investment to return. Meanwhile, the perceived cost of adoption, not just financial outlay but the time and effort required to retrain teams and adjust processes, acts as a further deterrent, particularly for firms operating on thin margins.
The construction sector’s long project timescales often serve to reinforce these problems, as outdated processes become embedded in live contracts and nobody wants to change how data is captured halfway through a job when compliance obligations have already been locked in.
Winning the trust
Taken together, the pros and cons of AI point to something subcontractors should welcome rather than fear. Proper use of this tech won’t eliminate risks, but it can empower businesses to find efficiencies and address blind spots. More complete verification, cleaner data, and real-time environmental awareness all add up to fewer disputes, lower risk, and more time spent on the work that actually pays.
While AI holds genuine transformative potential for subcontractors, that potential is only realised where the workforce can interrogate it, not simply defer to it. Technology without trust in the humans wielding it will ultimately be another tool gathering dust in the site cabin.
The post AI in construction: When it’s the right tool for the job – and when it isn’t appeared first on Planning, Building & Construction Today.
