The construction industry took the better part of a decade to put smartphones in workers’ hands. We don’t have a decade for AI in construction.
I’ve been running Assignar for twelve years. In that time, I’ve watched our industry move at its own pace — cautious, hands-on, prove-it-to-me. There are good reasons for that. Construction is physical. The stakes are real. You can’t undo a poured slab or a missed safety check.
But something has shifted in the last twelve months that’s different from anything I’ve seen since we started this company. But AI isn’t on the horizon for construction anymore. It’s already inside our customers’ businesses, doing real work, every day. And the contractors who treat it like a co-pilot — not a threat, not a chatbot, not a science experiment — are pulling away from the pack.
This is a point of view from someone who spends most of his time talking to construction operators across Australia, New Zealand, the US, Canada, and the UK. Here’s what I’m seeing, what I think it means, and what you should do about it.
AI in construction is no longer about chat. It’s about doing.
A year ago, “using AI” mostly meant typing a question into ChatGPT and getting a better Google answer. That was useful. But it wasn’t transformative.
Today, the contractors I talk to aren’t just chatting with AI. They’re handing it work. Real work. The kind of work that used to live on someone’s desk for two weeks while everyone else waited.
The behaviour change we’re hearing loudest from our customers right now: they don’t want AI sitting in a separate browser tab. They want it plugged directly into the systems their real work lives in. That’s why we built an MCP for Assignar — a way for our customers to connect Claude, ChatGPT, or whichever AI they’re using straight into their Assignar data and pull the insights that matter most to their business, in plain English. Worker utilisation, overtime hot spots, expiring tickets, project margin, EBA compliance flags, last week’s safety trends — whatever question the contractor is running their business on, the AI now answers it off live Assignar data. Customers are already using it. If you’re on Assignar, the MCP is live at mcp.assignar.com and our team can walk you through setup.
A few of the patterns I’m watching across our customer base:
Contract reviews in minutes, not days.
The days of one person ploughing through an 80- or 100-page head contract are ending. Contractors are loading those contracts into AI and asking, “What are my payment terms? Where am I exposed on retentions and variations? What do I need to submit with my payment claim? Which clauses should I be pushing back on before I sign? And what should I take to my lawyer?” The review happens in minutes. And because modern AI tools have memory, the AI becomes an expert in your business over time — not a blank slate every time you ask.
Safety insights from data you already have.
Most contractors have digitised their toolbox talks, pre-starts, and inspections. The forms get filled out. They sit in a database. Nobody has time to plough through a hundred form submissions to spot the leading indicator that matters. AI does. We’re in deep conversations with tier-one contractors right now about using AI to aggregate sub-contractor safety data across their job sites and surface the patterns that would otherwise get missed.
Quote automation for inbound work.
I had a conversation with a demolition contractor in Perth recently who said, “We get residential demolition enquiries every week off our website, and I lose track of responding. I have so much work, but I’m leaving quotes on the table.” His next step is straightforward: he’s going to set up an AI agent to read his inbox at 7am every Monday, pull the property address from the email, look up dimensions on Nearmap (because that’s how he prices the job), draft a quote in his quoting system, and have a reply ready for him to review before he sends it. He’ll have it done by the time he’s finished his coffee.
EBA and award interpretation.
Our customer KPI runs about a thousand workers nationally across eight to ten different EBAs. Those documents are long, dense, and unforgiving. Shay, one of KPI’s directors, told me at our Build Better Melbourne event that AI is now part of how her team interprets them. That’s high-stakes, highly-regulated work, and AI is helping them do it faster without losing accuracy.
None of this is theoretical. None of this is “coming soon.” It’s running, right now, in construction businesses that look a lot like yours.
The co-pilot mindset: an extra set of hands, not a replacement
A lot of people are nervous AI is going to take their jobs. In some industries, that may be true. I don’t think it’s true in construction any time soon — at least not in the way people fear.
Here’s the framing I keep coming back to: AI is a co-pilot, not a replacement. Think about it as an extra set of hands. Multiple sets of hands, actually. The grunt work, the admin, the data entry, the document review — the things you don’t necessarily need a human doing anymore — that’s where AI shines.
Use AI for better work, not more work
One trap to avoid: when AI gives you time back, it’s tempting to use it to do more — more reports, more emails, more requests pushed down the chain, more volume of everything. Resist that. The point of a co-pilot isn’t to crank out more of what you were already doing. It’s to lift the quality of the work you’re already doing. Better-thought-through quotes. Sharper contract reviews. Safety insights you actually act on. A pre-start that catches the issue before the operator gets in the cab. AI is at its most valuable when it helps you trade quantity for quality, not the other way around.
Meanwhile, the work that remains is the work that always mattered: judgement, relationships, decisions on the ground, the moments when someone needs to stand on a site and make a call. AI doesn’t do that. It frees you up to do more of it.
When I write software now, I don’t work with one AI agent. I work with swarms of agents — in parallel. One is a construction expert. Another is a civil engineering expert. A third covers safety. A fourth covers quality. They talk to each other. They check each other’s work. And there’s still a human (me, or one of our engineers) reviewing what they produce before it ships.
You can do the same thing in your business. There’s no need to be a software company, and no need for a tech team. The tools that let you run ten agents in parallel are available right now, off the shelf, for a price most contractors spend on coffee.
What this means for your software vendors
If AI in construction means software is being built at a fundamentally different pace, that should change what you expect from every software vendor in your stack — including us.
A few numbers from inside our own business:
- In 2025, we shipped 650 enhancements to the Assignar platforms. That was roughly three times what we shipped the year before, because we started using AI seriously in the back half of the year.
- In 2026, we’ve already shipped 230 enhancements by mid-May. Extrapolated linearly, that’s well over 900 by year-end. The actual number will be higher, because multiple agent teams now run in parallel.
- Our Assignar Pay platform took my co-founder Marko and I eight months to build to launch. Without AI, that same build would have taken closer to two years and a team of fifteen to twenty.
Now, I’m not sharing those numbers to brag. I’m sharing them because you should expect this from every vendor you buy from. If your software provider is shipping at the same pace they were two years ago, ask why. If they tell you “AI features are on the roadmap” with no measurable acceleration in product velocity, ask harder.
Pressure-test us. Pressure-test all of them.
A useful test for any software vendor you’re paying
Here’s a useful test: tell your vendor about a feature you’ve been asking for. If their answer is “it’s in the backlog” without a delivery date, that’s a 2023 answer. In 2026, the right answer should be measured in weeks, not quarters. We built a long-requested daily planning view inside Assignar Operations recently — drag-and-drop, full short-range visibility, the whiteboard-on-a-screen our customers had been asking for for years. Historically, that’s a four-to-five-month build with a team of five or six. We delivered it with one engineer in two weeks, fully tested, production-ready.
In short, that’s the bar now. The bottleneck isn’t engineering. It’s humans reviewing the work for safety, security, and quality — and that’s exactly where the human still belongs.
The next frontier: AI in the hands of your field workers
Most of the AI conversation right now is happening in the office. Contract reviews, quote drafting, dashboards, reporting. That’s the obvious starting point. The bigger shift — and the one that will reshape how construction businesses actually run — is AI in the field.
Think about how clunky software has been for our crews on a job site. Buttons. Forms. Required fields. Spell-check warnings. None of that is how the people doing the actual work want to communicate.
Modern AI tools have flipped the interface on its head. There are no buttons. Just a text box. Just a microphone. You talk to it like you’d talk to a coworker. And it figures out the rest.
Why voice changes the game for field workers
In construction, that’s a profound unlock. A lot of the men and women out on site are more comfortable talking than typing. They have decades of expertise that lives in their head and comes out in conversation, not in a structured form. If we can give them a way to capture that — verbally, in the moment, on a mobile device — we get richer data, faster reporting, and people who actually use the tools we built for them.
And that’s where we’re heading with Assignar Pay’s Field module. It’s also where the industry is heading, regardless of which platform you’re on. If your software roadmap doesn’t have voice and natural language interaction in it for field workers, your software is going to feel old very quickly.
A note on data privacy (because it always comes up)
Every time I talk about AI with a room of contractors, the first question is the same: “What happens to my data?” It’s a fair question. You’ve got sensitive worker information, signed NDAs, confidential client agreements. You don’t want any of that ending up in someone’s training data.
Here’s the short answer: use a business account, not a personal one. With Anthropic (the maker of Claude), OpenAI, and every other serious AI vendor, business plans let you opt out of having your data used for training. Get your legal team to review the contract. We did, before we rolled AI out across Assignar. It’s not bulletproof, but it’s a meaningful line.
And when you give an AI tool access to your files, give it access to a folder — not your whole system. Scope it. Treat it like you’d treat a new hire on day one: trusted, but not unsupervised.
Where to start (the practical bit)
If you’ve read this far and you’re thinking, “this all sounds great, but I don’t know where to begin” — here’s the most useful exercise I can give you.
List everything you do manually in a week. Every report you compile by hand. The emails you draft from scratch. The contracts you skim. Forms you build. Spreadsheets you reconcile.
Then go to one of the modern AI platforms (Claude, ChatGPT, pick your flavour), spend $20 a month on a business account, and ask it: “Here’s what I do every week. How can you help me automate or speed up any of this?”
You’ll be surprised how much of it the AI can take a real swing at. The trick is to give AI the work it’s genuinely better at than a human — the mundane, repetitive, admin-heavy stuff that drains your week. And don’t settle for chat-style back-and-forth. Use agentic workflows, where the AI has the context it needs (your contracts, your templates, your rate cards, your forms) and can take actions on your behalf: drafting the email, building the form, pulling the report. Then check the work when it’s done. That’s the rhythm — AI runs the task, you review the output. Start with one workflow. Get good at it. Then move to the next.
In fact, this is exactly what we’ve challenged our own team at Assignar to do. The wins compound.
AI in construction isn’t coming. It’s here.
The contractors I see pulling ahead aren’t the ones with the biggest tech budgets. They’re the ones with the most curious leadership — owners and ops leaders who treat AI like a coworker they’re onboarding, not a threat to ignore or a magic wand to overhype.
Still, construction has always rewarded the people who can move faster than the next contractor without dropping the ball on safety or quality. AI doesn’t change that. It just raises the ceiling on how fast “fast” actually is.
If you’re an Assignar customer and you want to see how we’re embedding AI inside Operations and Pay, talk to your account team. If you’re not, book a demo and we’ll show you what AI in construction actually looks like when it’s running underneath your scheduling, your timesheets, your forms, and your pay run.
The industry is moving. Don’t wait it out.
Sean McCreanor is the CEO and co-founder of Assignar, a construction operations and payroll platform used by contractors across Australia, New Zealand, and North America.
