Jack Dorsey Just Cut His Company Nearly in Half—Because AI Changed the Math.
Jack Dorsey—founder of Twitter, founder of Block—posted something on X this week that should make every CEO, tech or not, sit up straight.
Block is cutting its organization nearly in half. From over 10,000 people to just under 6,000. More than 4,000 people asked to leave or entering consultation.
And he didn’t blame a downturn. He said the business is strong. He pointed to something else:
“We’re already seeing that the intelligence tools we’re creating and using, paired with smaller and flatter teams, are enabling a new way of working which fundamentally changes what it means to build and run a company. And that’s accelerating rapidly.”
Now, you might be thinking: That’s a tech company. We build things. Steel, concrete, glass. What does Jack Dorsey have to do with me?
More than you think.
The shift he’s describing isn’t “tech.” It’s coordination. And if you’ve been in this industry for more than five minutes, you know that construction is a coordination business with a jobsite attached. We coordinate people, information, money, materials, and time. That’s most of the job. The actual building—the craft—is the part we’re good at. It’s the coordination that buries us.
I’m not writing this to scare you. I’ve been working with construction companies since 2004, and I’ve watched this industry survive the 2008 crash, the lean years that followed, and COVID. Construction leaders are tough. We just keep going.
But I’m a pessimist by nature. I’m the guy who looks at a profit and loss statement, skips the profit, and scans for the loss first. So when I see a tech CEO cut 4,000 people not because his business is failing but because the tools got that much better—I pay attention. And you should too.
Matt Blauch, president of Blauch Brothers Mechanical Contractors, put it to me this way after going through our Construction Genius ChatGPT Course: “AI isn’t optional anymore—it’s reshaping the business landscape. We’re watching major companies achieve massive efficiencies through AI, with some reducing their workforce by tens of thousands. The mindset that ‘AI won’t affect us’ is dangerously shortsighted.” Matt’s a mechanical contractor, not a tech executive. And he sounds exactly like Jack Dorsey.
The construction CEOs who think clearly now and start making moves now will be the ones leading in 2029. The ones who wait until it’s obvious will be reacting instead of steering.
So let’s walk through the four things your business does every day: winning work, planning work, building work, and getting paid. Here’s what’s happening right now in each one, and where it’s heading.
1) Bidding Work: The Estimating Advantage Is Shifting
For most construction companies, estimating is a serious competitive advantage. You’ve got people who know your market, your costs, your subs, and how to sharpen a number to win without bleeding margin. That knowledge has real value.
But the tools are changing fast. And some of your competitors are already using them.
AI Tools Already Changing How Construction Companies Estimate
AI-assisted takeoff is real. Procore’s estimating platform uses computer vision to detect floor plan areas and count repeated symbols in plan sets—work that used to eat up a skilled estimator’s afternoon. Their Helix intelligence layer, launched at Groundbreak 2025, adds AI-assisted insights inside the platform as teams build bids.
Risk scanning before bid day is here, too. Tools like Provision apply AI to plans, specs, and related project documents to surface scope gaps, quantity anomalies, and potential conflicts before you submit. That’s the kind of omission that turns into a change order six months into a project. Now there’s software catching it on the front end.
And the specs-to-submittals process is getting automated. Autodesk’s preconstruction tools use AI to surface potentially missing submittal items by comparing your specifications against historical project data. That used to require a PM manually grinding through spec books page by page.
These tools don’t make decisions for you. They reduce blind spots and compress time. Think of them like a second set of eyes on your estimate—eyes that don’t get tired at 2 AM before bid day.
I’m already seeing this play out with the construction companies I work with. Grant Gourley, SVP and COO at WM Lyles, told me that after going through our ChatGPT course, his team saw immediate results in spec review and data search. Tasks that used to take thirty minutes to an hour—reading specs, drafting letters, reviewing client communications—now take minutes. And Christina Phillippe, CFO at Urata & Sons Concrete, reported that one of her project managers used AI to review specifications and consolidate key facts, cutting several hours of work down to twenty minutes. That kind of time savings in preconstruction adds up fast when you’re chasing multiple bids.
The Estimating Edge in 2029: Systems Beat Individuals
The companies that build AI into their estimating process now are going to pull ahead. They’ll chase more opportunities with the same headcount. They’ll turn bids faster. And they’ll catch more quiet mistakes before those mistakes become margin-killers.
Here’s the harder truth. In three to five years, the estimating edge won’t be who has the best estimator. It’ll be who has the best estimating system—and who trained their people to judge AI output intelligently instead of just producing input.
The risk isn’t that AI replaces your estimators. The risk is that a competitor with half your estimating overhead starts beating you on speed, price, and consistency because their process is tighter. That’s what happened in every other industry where the tools got better and some companies adapted faster than others.
Question to ask yourself today: If your top estimator walked out tomorrow, how much institutional knowledge walks out with them? What are you doing right now to capture and systematize that knowledge?
2) Planning Work: From Reactive Schedules to Living Systems
Construction planning is managing chaos. Weather. Lead times. Manpower. Inspection windows. Owner decisions. Trade stacking. One late submittal turns into a cascade that blows your schedule and eats your margin.
Most schedules are best guesses that start going stale the moment the job begins. You know this. I know this. We’ve all lived it.
The tools being built right now don’t just make planning faster. They push it toward being continuous.
How AI Is Replacing Stale Schedules with Real-Time Project Intelligence
Autodesk Construction IQ uses machine learning to scan your project data—RFIs, submittals, issues, checklists—and prioritize what’s most likely to become a cost or schedule risk. Autodesk reports it was used over five million times last year. That’s not a pilot program. That’s adoption at scale.
ALICE Technologies runs thousands of sequencing and resource scenarios before you break ground, then helps teams evaluate trade-offs between duration, cost, and labor. ALICE reports average reductions of 17% in project duration and 14% in labor costs on complex projects. Take vendor numbers with a grain of salt, sure—but the direction is clear.
And here’s one that’s practical right now. Trunk Tools built TrunkText so your superintendents can text questions from the field and get answers grounded in actual project documents—in seconds. No digging through a trailer full of plan sets. Suffolk Construction has been piloting it on live projects, and their supers are among the heaviest users. Think about what that means for a guy in the field who needs an answer at 6:30 AM and nobody’s in the office yet.
This kind of shift is already happening inside the companies I coach. Dimitri Sidiropoulos, VP at Delphi Construction, told me his team built a JHA bot after going through our ChatGPT course. It reads the scope of work, pulls the OSHA construction 1910 standards, identifies hazards, and lists mitigations. Once it’s fully built out, it’ll save around five hours per job for every JHA submission. Five hours. Per job. And that’s one application. Dimitri also dropped a messy two-hour operations meeting transcript into ChatGPT and got four pages of recommendations that could cut the meeting to thirty minutes. In ten minutes. He said it took what was in his head and got it onto paper fast.
Fewer Layers, Better Decisions: What Project Management Looks Like in 2029
The project executive of 2029 won’t be reading reports that are 48 hours old and trying to figure out what’s actually happening on the job.
They’ll have something closer to a living view of cost, schedule, risk, and cash. Not because you hired more people to compile reports. Because your people stopped spending half their day collecting and formatting information and started spending it making decisions.
This is what “smaller and flatter teams” looks like in construction. Not layoffs as a strategy. Better leverage from every leader you’ve got.
A senior PM who today needs a swarm of people just to manage information flow may be able to handle the same complexity with fewer layers—if your workflows and data discipline are strong.
Question to ask yourself today: Are your project managers spending most of their time gathering and organizing information, or making decisions? Because gathering and organizing is the part being automated first.
3) Building Work: What Changes Around the Craft
The field is the last place most construction CEOs expect AI to matter. You can’t automate a formwork crew. You can’t send a chatbot to install mechanical systems.
The craft is the craft. That’s not changing. But everything around the craft is.
AI in the Field: Daily Reports, Safety Docs, and Reality Capture
Daily reporting is being compressed. Tools like Raken combine photos, crew counts, weather, and quick inputs—including dictation and voice capture—so your superintendents aren’t stuck doing a second shift of paperwork at night. You know the problem. The super gets home at 7 PM, eats dinner, then sits down to type up daily reports for another hour. That’s a burnout machine. These tools cut that time significantly.
AI-assisted organization of field data is emerging. Platforms like InspectMind and StruxHub aim to take messy inputs—voice notes, photos, quick observations—and turn them into structured reports, tagged issues, and routed workflows. This space is moving fast. I’d evaluate any individual vendor carefully before signing a contract. But the direction is real.
Reality capture is here too. Buildots uses 360° cameras to document site conditions, then applies AI to compare progress against the model and the schedule—flagging discrepancies and tracking trade progress automatically.
And here’s the angle most CEOs miss: documentation is becoming a competitive weapon.
Major disputes and delay claims are won or lost on contemporaneous records. What you documented. When you documented it. How consistently you did it. Look at Walsh Construction v. Toronto Transit Commission, a 2024 Canadian case. Contemporaneous documentation played a central role, with expert analysis supporting over 1,000 days of compensable delay. The takeaway is simple: clean, timestamped records win fights. Contractors using voice-to-text and AI documentation tools today are building that kind of evidence base as routine. The ones still on paper logs are not.
Charlie LaMendola, Director of Construction at Schultz Construction, told me that AI has become his go-to tool for exactly this kind of work. He deals with claims, technical engineering questions, and owner correspondence on a daily basis. “Where I used to spin my wheels trying to get answers,” he said, “AI has become extremely handy.” Charlie came into our ChatGPT course using AI like most people—as a glorified Google search. Now he uses it for research, correspondence, and the legal and contract side of his work. That’s the shift. It’s not about the tool being impressive. It’s about the tool saving a construction professional real time on real work.
The Field Leadership Challenge No One’s Talking About
Your best superintendents—the ones who are gifted at reading the job, managing trades, and solving problems in real time—are going to be freed from a chunk of the administrative burden that burns them out.
There’s long-cited industry research from FMI, PlanGrid, and others showing construction professionals spend a massive share of their time on non-productive activities: searching for information, dealing with conflicts, correcting errors. You don’t need the exact percentage to know it’s true. Your people feel it every day.
Cut that meaningfully, and your field leaders can manage more scope with the same energy. That’s huge when you’re already struggling to find and retain good field leadership.
But here’s the leadership challenge coming faster than most CEOs expect. Field leaders who refuse to adapt—who treat these tools as a nuisance or a threat—will become a liability. Not because they’re bad at their jobs. Because competitors whose teams run lean, documented, and fast will eat their lunch. Developing your field leadership for this reality isn’t an HR initiative. It’s a competitive strategy.
Question to ask yourself today: What percentage of your superintendents’ time is spent on tasks that add zero value to actually building the building? What happens to your project economics if you cut that in half?
4) Getting Paid: The Last Manual Process in Your Business
If there’s one area of construction that still feels stuck in the 1990s, it’s billing and collections.
Pay apps. Lien waivers. Retainage. Change order backup. Compliance. Owner review cycles. Rejected apps. Rework. Resubmittals. It’s an avalanche of paperwork that delays your cash and creates disputes. And for most companies, it’s still almost entirely manual.
Pay Apps, Lien Waivers, and the AI Tools Automating Cash Flow
Pay app workflow automation is already in active use. Platforms like GCPay and Trimble Pay automate key parts of pay application collection from subs, lien waiver compliance tracking, and integration into your accounting and ERP workflows—so data doesn’t get re-entered by hand. These aren’t experimental tools. They’re working right now.
AI-assisted review is emerging. Vendors are building tools that review pay app packages—including common AIA forms—for consistency and errors, cross-verifying totals against prior payments and continuation sheets, then flagging issues before they trigger rejection and delay. This is early-stage, but the potential to save time and reduce the back-and-forth is obvious.
And the broader direction is to tie billing to live field progress. Some platforms are building toward tighter linkage between what’s captured in the field and what goes into the pay application—so your numbers flow from the job instead of being assembled manually at month-end. Adoption is early, but the concept makes sense.
Here’s one you might not expect. Matt Blauch at Blauch Brothers told me the ChatGPT course taught his team how to compare WIP reports using AI. But then he realized the same principles applied to analyzing their monthly credit card statements—often totaling $150,000. What previously took hours of manual review—identifying top vendors, categorizing non-fuel spending, flagging items for review—now takes minutes. That single application, which wasn’t even covered in the course, justified the entire investment. That’s what happens when your people understand how AI actually works. They start finding applications you never anticipated.
Why the Firms That Fix Billing First Will Have a Cash Flow Advantage for Years
Companies that systematize billing and collections will have a cash flow advantage. Cleaner submissions. Fewer disputes. Faster approvals. Better visibility into what’s owed and when.
But I want to be honest about this one. Getting paid isn’t just a technology problem. It’s a contractual and legal ecosystem involving owners, lenders, bonding companies, and sometimes government agencies. Broad adoption across the whole industry will lag the other three areas. Probably closer to a five-to-seven-year curve before it’s standard practice.
That said, the firms that start now will be years ahead of the ones waiting for it to become universal. Cash is oxygen. You know that. The number one reason construction businesses fail is cash flow. Anything that tightens the cycle between doing the work and getting paid for the work is worth your attention.
Question to ask yourself today: How many days does it take from the time work is completed to the time you submit a pay app? How many more days to collect? What would even a 20% improvement do for your cash and your stress?
The Leadership Lesson Jack’s Decision Is Really About
Let me come back to Dorsey, because the most important thing he said wasn’t about technology. It was about decision-making.
He said he had two options: cut gradually over months or years, or act decisively now. He chose the hard, clear move. Because, as he put it, repeated rounds of cuts destroy morale, focus, and trust.
Most construction CEOs will instinctively choose the gradual approach. Not out of weakness. This industry values relationships, loyalty, and steady management through cycles. That culture has served construction well.
But what’s coming isn’t like the cycles you’ve managed before.
From 2004 to 2007, we were all geniuses. Then 2008 hit and for the next four years we were all idiots. Then the money came back and we were geniuses again. Then COVID. I’ve watched this industry ride these waves since the beginning of my career.
This is different. This isn’t a recession. This isn’t a supply chain spike. This isn’t a labor shortage. Those are cyclical. They come and go.
AI is structural. It’s a shift in the cost of coordinating complex work. And it doesn’t reverse when the economy turns.
The leaders who will be strongest in 2029 are the ones asking hard questions now. Not “Should we buy some AI software?” but: What does our business model look like when AI reduces our administrative overhead by 30 or 40 percent? Who do we need then? What do we charge for? What is our actual competitive advantage when the tools are available to everyone?
I’ll tell you what I’m seeing with the companies I work with. The biggest shift isn’t technical. It’s cultural. Christina Phillippe at Urata & Sons Concrete said the same thing: “The biggest shift has been cultural—people are more open to AI and see its value. For those who were initially skeptical, the course helped demystify the technology and show how effective it can be when used thoughtfully.” Grant Gourley at WM Lyles is already building an AI training module into their company education program. Charlie LaMendola told me flat out: “If the course had gone on for another few months with more content and topics, I would have stayed and kept doing it.” These aren’t tech enthusiasts. These are construction professionals who got their hands on the tools and realized this changes how they work.
Most firms are still early. The differentiator won’t be buying AI. It’ll be changing your workflows and cleaning up your data so AI can actually help. It’ll be developing your people to work with these tools instead of fighting them. It’ll be making the hard calls about where your time and money go.
That’s leadership. It always has been. The tools change. The job of the leader doesn’t.
Own your time, own the future. And the future just got a lot closer.
Get Your Team Ready Before Your Competitors Do
The construction executives quoted in this article—Matt, Grant, Christina, Dimitri, Charlie—all went through the Construction Genius ChatGPT Course. They came in with the same skepticism you probably have right now. They left with their teams using AI on real work, saving real hours, every week.
We run the course in small cohorts with company-specific breakout rooms, so your team works on your problems, not generic examples. The next cohort is forming now.
If you want your people to stop dabbling and start getting results with AI, get on the waiting list. When the next session opens, you’ll be the first to know.
➤ Join the Waiting List for the Next Construction Genius ChatGPT Course
Click here to join the waiting list
Spots are limited. We keep the cohorts small so every team gets direct facilitator support. If you’re serious about getting ahead of this, don’t wait until your competitor’s team is already trained.
