19/10/2026
AI in construction is stalling. The reasons might surprise you.
Bluebeam’s AEC Technology Outlook 2027 surveyed a heap of US architecture, engineering and construction decision-makers. For us, it highlighted some interesting points! Only 7% have AI embedded across multiple core workflows, and 32% are stuck in pilots.
(We'll get into how that translates for us down in NZ and Australia shortly.)
So why is AI stalling in bigger construction markets? Unsurprisingly, some familiar sounding issues...
🚧 No clear job for the AI, and messy data are a big issue. Stalled firms were far more likely to be stuck on what to build and whether their data could support it. A pilot with no clear purpose has nothing to scale.
🚧 The systems don’t talk to each other. Work breaks down where it passes between people and tools, especially between the office and the field. AI added to one tool doesn’t fix that.
🚧 Cost and skills worried every firm, including the ones that scaled.
⁉️ Where cost predictability fits
Cost predictability was the top benefit firms reported from AI. That makes sense. AI is good at learning from office data like past quotes, job costs and schedules, so it can forecast a blowout earlier.
But rework, which is what actually eats margin, was near the bottom of the benefits. Stopping a blowout needs live information from site reaching the people who can act on it, and for most firms that link is missing. AI can predict the problem, but it can’t prevent it without current field data.
⁉️ What this means for Kiwi & Aussie firms
Most of our trades and builders are small, with no IT team and no spare time to untangle new tech. That’s where the right job management system fits. Tracking quoted vs. actual (what you priced a job at against what it’s really costing) is only useful if it’s current. In systems like WorkflowMax (ConTech are Platinum Partners) that can be a flag when a job has burned 70% of budget at 40% complete, while it’s still fixable.
With MCP server updates in the advanced package, you can also ask your system questions in plain language, like you would with Claude or ChatGPT. So, you get the best of both worlds. A bit of AI but a sold back end.
If you need an easier better way for your field team to enter in the information remotely so your business can make better decisions with cost predictability, it works for that too. We've even done custom development for our clients to simplify this further, so there's never a "too many cooks in the kitchen" type scenario that can (if unchecked) make your data go a bit wonky.
At ConTech, “AI-ready” means the unglamorous stuff: consistent job codes, timesheets that get filled in, and quotes you can compare with actuals. With Master Builders NZ warning that a shortage of work could flip to a shortage of workers within 12 to 24 months, the quiet patch is the time to get set up with smarter systems. We agree!
Where does your job data stand today? Are you using AI? What are your thoughts on what works, and what doesn't? We'd love to hear from you in the comments.
