The Tool Changed.
The Expertise Didn't.
The construction industry has spent decades waiting for technology to catch up to the people who actually build things. It finally did — just not the way anyone expected.
Cain Menard
Director of Consulting & Operations · 18 min read
Remember that GEICO commercial? A couple of well-groomed cavemen sitting in a nice restaurant, visibly offended that GEICO just told the world their website was “so easy, a caveman could do it.”
The joke worked because everybody assumed some people just aren't the tech type. Smart, sure — just not that kind of smart.
That's over. The tools changed. And the people who know the most about how construction actually works are about to become the most valuable players in the room — not despite their lack of technical background, but because of the expertise they already have.
Key Takeaways
- AI coding tools reduced a weeks-long Tableau dashboard rebuild to hours as a fully deployed React web application
- Domain expertise — not coding ability — is the new bottleneck for AI adoption in construction
- Construction productivity has grown only 10% since 2000 vs. 90% in manufacturing
- 96% of construction data goes unused — the industry is the 2nd least digitized in the U.S.
- Organizations with structured change management are 7x more likely to succeed with new technology
- The value is shifting from software itself to the expertise of knowing what to build
- Readiness — people, process, data, integration — matters more than the tool
The Shift
The Gap Between Thinking and Building Just Collapsed
A completed project performance dashboard built over several weeks or months three years ago — data modeling, formulas, iterative design, testing, troubleshooting, and revisions — can be rebuilt as a full production web application in just hours today. 5 powerful dashboard views, interactive charts, KPI cards, cross-dimensional filters, executive insights, and strategic recommendations. Deployed for free. No license, no login, no IT department involved, using Claude Code as an AI pair-programmer and Wispr Flow for voice-to-code.
The process was stream-of-consciousness — describing what the application should do in a long, ranting “word salad,” refining in real time, watching production-quality lines of React code materialize from verbal instructions.
It's not a prototype. It's a deployed application, built by describing what it should do in plain English.
The Proof
What Changed
Same analytical goal. Same domain expertise. Radically different process and outcome. Both tools below are live and fully functional — click through and use them.
The Evidence
The Numbers Behind the Shift
This isn't an edge case. The productivity data is accumulating fast.
GitHub's controlled experiment found developers using AI coding tools completed tasks 55.8% faster.1 Harvard and BCG tested 758 consultants and found AI users produced 40% higher-quality output while finishing 25% faster.2 A Google principal engineer gave Claude Code a three-paragraph problem description and received a working result in one hour that matched what her team had spent a year building.3
Google now generates roughly half its new code with AI.4 Over 90% of Fortune 100 companies have deployed GitHub Copilot.5 Gartner projects 90% of software engineers will use AI code assistants by 2028, up from under 14% in early 2024.6 The AI code tools market is valued at $7.4 billion and heading toward $25–30 billion by 2030.7
Andrej Karpathy, co-founder of OpenAI, coined what he calls “vibe coding” and declared: “The hottest new programming language is English.”8 Jensen Huang, NVIDIA's CEO, at the World Government Summit: “Everybody in the world is now a programmer.”9 Satya Nadella, in his 2025 letter to shareholders: “More than any transformation before it, this generation of AI is radically changing every layer of the tech stack.”10
For balance: a rigorous 2025 METR study found experienced open-source developers were actually 19% slower with AI tools on complex maintenance tasks11 — while believing they were 20% faster. AI excels at building new things from domain knowledge. It struggles with maintaining complex legacy codebases. That distinction matters significantly for how companies should think about adoption.
The Real Builders
The Real Builders
The Expertise
Superintendent
Knows which crews fall behind before the schedule shows it
Estimator
Knows which cost drivers eat margins before the bid closes
Project Manager
Knows which indicators flag trouble six weeks before the financials
What They'll Build
Crew Performance Tracker
Real-time field monitoring with automated alerts
Cost Analysis Dashboard
Live margin tracking and variance reporting
Project Health Monitor
Early warning system for at-risk projects
Let me be clear — superintendents, project managers, and estimators aren't cavemen. They're the people who actually understand how projects make or lose money. Which cost drivers eat margins. Which reports are just busy work nobody reads. Which leading indicators flag trouble six weeks before it shows up in the financials. That's fifteen, twenty, thirty years of hard-earned knowledge that no software developer in Silicon Valley is going to replicate.
The problem has never been that construction professionals aren't smart enough for technology. The technology has never been smart enough for them. Good tech adapts to the user, and the construction industry has been forced to settle for less. I wrote about this a year ago: software developers should be upskilling and adapting their work to serve the construction industry — not the other way around. I still believe that.
Until now, the IT department — if the company even had one — occupied a small, dark corner of the office. JBKnowledge found fewer than half of construction companies have a single dedicated IT employee.12 I still walk into companies where field supervisors don't have devices. Paper daily reports. Handwritten timekeeping. Multi-million-dollar projects managed with the same information tools available in 1985.
Meanwhile, those same companies are struggling to hire younger talent born into a digital-first world. Gen Z participation in construction more than doubled between 2019 and 2023.13 But 41% of the pre-2020 workforce is expected to retire by 2031.13 There's a ticking clock on the institutional knowledge in your people's heads — and no clear mechanism to capture it before it walks out the door.
Here's the shift most people miss: those same superintendents who've never written a line of code may become your most prolific software builders. Building custom applications is becoming as simple as describing what you need out loud, the way you'd explain it to a colleague.
Kevin Roose, a New York Times technology columnist, built a functional app in about ten minutes.14 A non-coder in the Philippines built a custom expense management app in two hours using plain language prompts. Harvard Business School has an active teaching case on it — “Lovable: Vibe Coding for the Other 99%.”14
If those people can do it, your best PM can do it. And the software they'd build would actually solve the problems they deal with every day — because they're the ones who understand those problems.
The Industry
The Industry That Stands to Gain the Most
Productivity Growth, 2000–2022
Indexed output per worker
Construction productivity grew 10% between 2000 and 2022. Manufacturing grew 90% over the same period.15 Let that sink in. And from 2020 to 2022, construction productivity didn't just stagnate — it actually declined by 8%.15
Construction is the second-least digitized major industry in the United States.16 Most companies spend less than 1% of revenue on IT.16 For comparison, automotive and aerospace spend 3–5%. The gap is massive — construction underspends cross-industry IT averages by 60–70%.
And the tools that do get implemented? They rarely deliver. Over 70% of ERP implementations fail to meet their original business objectives.15 The industry hemorrhages an estimated $177 billion a year to operational inefficiencies.15 In 2020, poor data management alone cost the construction industry $1.84 trillion globally — and only 55% of companies even have a formal data plan.17 Ninety-six percent of the data generated on construction projects goes completely unused.17 Field teams lose nearly two full working days every week — 14+ hours — just searching for project information18 and dealing with problems that shouldn't exist.
None of this is a technology problem. It's a fit problem. The people who understand the work best — who know exactly what dashboard they wish they had, which reports are useless, which metrics actually predict when a job is going sideways — have always been the furthest from the tools that could help them. They've never had a way to turn that knowledge into software without a six-figure budget and a year-long timeline.
AI changes that math. The AI-in-construction market sits at $3.9 billion today, projected to hit $22.7 billion by 2032.19 Even Procore — the biggest construction management platform in the market — launched an Agent Builder that lets construction professionals customize AI workflows without writing a single line of code. The AGC's 2025 Workforce Survey shows 44% of firms already expect AI and robotics to improve job quality and productivity.13
The industry that's historically been last to adopt technology may have the most to gain from what's happening right now. The gap between what construction professionals know and what they've been able to build with that knowledge is wider here than anywhere else. That gap is closing fast.
The Pattern
The Dot-Com Parallel
The Dot-Com Pattern
Technology adoption follows a predictable arc
Source: Carlota Pérez, Technological Revolutions and Financial Capital
Not convinced? That's fair. The investment numbers warrant skepticism.
AI venture capital hit $202 billion in 202520 — capturing roughly half of all global VC funding. NVIDIA's stock rocketed 2,000% from 2022 lows to a market cap north of $5 trillion. Michael Burry called the AI boom “a glorious folly” and compared NVIDIA to Cisco, which surged 3,800% before crashing 88% and never recovering to its inflation-adjusted peak. Jeff Bezos called it “kind of an industrial bubble.” Sam Altman — ChatGPT's version of Steve Jobs — himself said, “People will overinvest and lose money.” An MIT study found 95% of 52 organizations achieved zero ROI from generative AI investments.21
“95% of organizations studied achieved zero ROI from generative AI investments.” — MIT Sloan22
The dot-com crash wiped out $5 trillion in market value.23 Pets.com went from IPO to liquidation in 268 days. Webvan raised over a billion dollars and was dead in two years.
But the technology survived. Internet users grew from 361 million in 2000 to over 4 billion — continuously, straight through the bust and out the other side. More than 70% of Americans were already online when the market bottomed out. The bad business models died. The technology didn't.
And the failed ideas came back with better execution. Webvan became Instacart. Pets.com became Chewy ($8.7 billion IPO). Kozmo became DoorDash. Amazon crashed from $107 to $7 a share before becoming one of the most valuable companies on Earth.
Carlota Pérez,24 a techno-economist who's studied five major technological revolutions, found they all follow the same arc: eruption, speculative frenzy, collapse, then a prosperous golden age. The question for construction leaders isn't whether there will be a correction. It's whether your company is positioned to come out the other side — because that's where the real money is made.
“The biggest and most sustainable profits tend to be made after the bubble has collapsed, not during the speculative frenzy.” — Carlota Pérez24
The Reckoning
Legacy Software Must Adapt or Disappear
Custom applications built by domain experts will begin replacing large portions of what legacy software currently does. Vendors that don't fundamentally enable deep configuration and customization beyond what's traditionally offered will lose market share. Some will cease to exist.
Microsoft CEO Satya Nadella declared “SaaS is dead” in late 2024.25 His argument: business applications are databases with business logic bolted on. In an agent-driven world, that logic migrates to AI agents that are database-agnostic. The application layer collapses.
The market is responding. The iShares Expanded Tech-Software Sector ETF fell over 23% in early 2026.26 Salesforce and Workday each dropped over 40% in twelve months.27IDC concluded SaaS has become “a patchwork of interfaces and data silos, forcing users to adapt to the software rather than the other way around.”28
IDC Prediction
By 2028, 70% of software vendors will need to fundamentally restructure their pricing models.28
Companies That Failed to Adapt
Kodak
Invented the digital camera in 1975. Suppressed it.
Bankrupt 2012
Blockbuster
9,000 stores, rejected Netflix.
Bankrupt 2010
Nokia
40%+ global market share. Missed touchscreens.
Sold 2014
Siebel
Created CRM. Failed cloud transition.
Acquired by Oracle 2006
In construction specifically, McKinsey found individual teams routinely build their own digital solutions without coordinating16 — creating a proliferation of competing, overlapping tools within a single company. JBKnowledge's surveys found 65% of respondents use spreadsheets for estimating12 despite having dedicated estimating software. The off-the-shelf products simply don't meet actual needs.
Every company has its secret sauce. Every contractor knows a good project from a bad one. Every business has different requirements. For the first time, there's an accessible pathway to building exactly what it needs — facilitating its functional processes, its data, and serving its people — without a software development team or a vendor's interpretation of “what construction companies want.”
The Bottleneck
Functional Experience Is the New Bottleneck
MIT labor economist David Autor argues that AI's unique opportunity is to “extend the relevance, reach, and value of human expertise.”29 That framing matters. Technical execution is being commoditized at a pace that would have been unimaginable five years ago. What's becoming scarce — genuinely scarce — is the domain knowledge that determines what to build and why it actually matters.
Erik Brynjolfsson at Stanford studied over 5,000 workers and found AI increased productivity 14–15% on average.30 But the breakdown is what's interesting: novice workers improved 34%, while experienced workers saw minimal gains. The AI was essentially encoding top performers' best practices and distributing them to everyone else. A study in Management Science made the point explicit — AI creates the greatest value when domain experts themselves can apply it, not when it gets filtered through an IT specialist who doesn't know the work.
“AI's unique opportunity is to extend the relevance, reach, and value of human expertise.”
— David Autor, MIT29
Here's where it gets interesting, though. That same Harvard/BCG study found that when consultants applied AI to tasks outside its capability boundary — things that require real judgment, intuition, or knowing which number to trust when two reports disagree — they were about 20% less likely to get the right answer31 than consultants working without AI at all. The tool made sloppy work worse. It exposed people who were mailing it in. Knowing when and where to apply AI was the single biggest differentiator between winning and failure.
That should matter to every construction leader reading this. Knowing how projects fail — budget overruns, material price escalations, change orders stacking up and eating your margin before anyone flags it — doesn't become less important in an AI-enabled world. It becomes more valuable, because for the first time that knowledge can be expressed as fine-tuned tooling rather than shoehorned into a vendor's template that was built for a generic version of your business.
Historical Parallel
The Tyranny of Spreadsheets
1980
339,000
accountants when VisiCalc launched
2022
1,400,000
accountants
The spreadsheet didn't replace accountants. It made them more productive and expanded the market.
Source: Bureau of Labor Statistics, via Tim Harford / Financial Times
The historical parallel is worth sitting with. When VisiCalc launched the first digital spreadsheet in 1980, there were 339,000 accountants in the U.S. By 2022, there were 1.4 million.32 Tim Harford at the Financial Times put it plainly: the spreadsheet didn't kill the profession. It made accountants more productive and blew the market wide open for what they could do.
Readiness
Readiness Comes Before Tools
Most construction companies aren't positioned to reap the benefits of AI tools yet. The problem isn't an inability to change or disinterest in finding better ways to work — every contractor has a story about “that one project”-turned success story thanks to a game-time judgment call that just might work. The problem is they haven't laid the foundation yet.
A peer-reviewed study of highway construction and asset management technology implementations found less than 10% of failures result from technical problems. Eighty percent of success depends on addressing people and process issues.33
“Any attempt to implement technology that focuses solely on technology is likely to fail in the construction industry.”33
Prosci's research shows organizations with structured change management are 7x more likely to achieve project objectives.34 A study in the Journal of Information Technology in Construction analyzed 167 technology-adoption cases across AEC firms35 and identified the top practices for success: change-agent effectiveness, measured benchmarks, realistic timeframes, and communicated benefits. Not better software. Not bigger budgets. People, process, and communication.
Readiness Framework
Talent
Do you have domain experts willing to learn new tools — and leadership willing to invest in them?
Process
Can you automate what you have, or are you automating chaos?
Data
Is your data clean, integrated, and accessible — or siloed and unreliable?
Integration
Do your systems talk to each other, or are they islands?
Process before technology. Data before dashboards. People before platforms.
Before any company can capture value from this technology shift, it needs honest answers to hard questions across four areas:
Talent. Domain experts willing to learn new tools, and leadership willing to invest in their development. With 94% of firms struggling to fill positions13 and mass retirements on the horizon, institutional knowledge is the most valuable and most perishable asset a construction company owns.
Process. Automating a broken process produces broken results faster. If an estimating workflow lives in fourteen spreadsheets with no version control, AI won't fix it. The Carnegie Mellon Capability Maturity Model36 makes the principle clear: the quality of any system is directly related to the quality of the process behind it.
Data. Ninety-six percent of construction data goes unused.17 Gartner predicts organizations will abandon 60% of AI projects unsupported by AI-ready data.37 The average construction business uses 11 separate data environments17 (often in the form of 11 disconnected systems). If data is siloed, duplicated, and unreliable, no tool will produce reliable outputs.
Integration. Nearly a third of construction companies report their systems don't communicate with each other.18 Disconnected estimating tools, PM platforms, accounting systems, and field apps don't just leave value on the table — they actively create the conditions for inefficiency.
AI coding tools are doing the same thing for domain expertise. The difference is that the industries with the widest gap between what people know and what they've been able to build with that knowledge stand to win the most — and by that measure, construction isn't just in the game. It's sitting on the biggest opportunity in the room.
Value
Where the Value Lives Now
The cost of building software is nose diving. Custom applications will proliferate. In the near future, domain experts across every industry will build their own tools fine-tuned to their specific operational processes.
That means the value of software itself — the code, application, and product — is in structural decline. Fewer startups will emerge to sell generic solutions at per-seat pricing. The market is already punishing that model.
What's not in decline: the expertise to know what to build, the operational knowledge to implement without disrupting what works, the ability to assess whether an organization is ready for new technology, and the strategic thinking that turns a compelling demo into measurable ROI.
“The barrier between knowing what you need and having it built has effectively disappeared. The question is no longer ‘Can we build it?’ It's ‘Do we know what to build — and are we ready to use it?’”
The companies that win in this next era will get the fundamentals right before chasing the fancy new tool. Process before technology. Data before dashboards. People before platforms.
The barrier between knowing what you need and having it built has effectively disappeared. The question is no longer “Can we build it?” It's “Do we know what to build — and are we ready to use it?”
Frequently Asked Questions
Common Questions
Can AI replace tools like Tableau for construction analytics?
AI coding tools can rebuild Tableau dashboards as full production web applications in hours rather than weeks. The resulting apps are free to deploy, require no special software licenses, and are accessible to anyone with a browser. However, AI excels at building new applications from domain knowledge — a 2025 METR study found experienced developers were actually 19% slower with AI on complex maintenance tasks.
How long does it take to build a web app with AI coding tools?
Build times vary by complexity, but the case study documents a complete project performance analytics dashboard — with 5 interactive views, KPI cards, cross-dimensional filters, and executive insights — rebuilt as a deployed React web app in hours using Claude Code and voice-to-code tools. GitHub's controlled experiment found developers using AI coding tools completed tasks 55.8% faster overall.
What skills do construction professionals need to use AI effectively?
Domain expertise is the most critical skill. MIT labor economist David Autor argues AI's unique opportunity is to extend the relevance and value of human expertise. Stanford research found AI creates the greatest value when domain experts themselves apply it, not when filtered through IT specialists. Construction professionals who understand project performance, cost drivers, and operational workflows are uniquely positioned — the barrier is now describing what to build, not knowing how to code.
Is the construction industry ready for AI adoption?
Most construction companies need foundational work before capturing AI value. Research shows less than 10% of technology implementation failures are technical — 80% depend on people and process. A readiness framework should address four pillars: Talent (domain experts willing to learn), Process (standardized workflows, not chaos), Data (clean, integrated, and accessible — 96% of construction data currently goes unused), and Integration (connected systems, not islands). Organizations with structured change management are 7x more likely to succeed.
What is the ROI of AI in construction?
The AI-in-construction market is projected to grow from $3.9 billion to $22.7 billion by 2032. However, an MIT study found 95% of organizations achieved zero ROI from generative AI investments — largely due to readiness gaps, not the technology itself. Construction loses an estimated $177 billion annually to operational inefficiencies, and field teams lose 14+ hours per week searching for project information. Companies that address process, data, and change management foundations first stand to capture significant returns.
Contact
Let's Talk
Whether you're a contractor trying to figure out why you're leaving money on the table, a firm looking to modernize operations, or just someone who wants to talk shop — I'm always up for a conversation.
(337) 654-2304 · Atlanta, GA