The AEC/O industry doesn’t need to be impressed. It needs AI that meets it where the work already happens — and more than 2,000 firms just showed the Nemetschek brand Bluebeam what that looks like.
Before we launched Bluebeam Max, our premium offering that is powered by AI and expands the power of Bluebeam Revu, more than 2,000 firms used it in beta. I’d like to tell you we handed them a finished product. What really happened is closer to the reverse: they handed us an education. Most of that education was about what doesn’t work. In construction specifically: telling builders their workflow is wrong. Telling them the new tool is the future. Telling them anything, really, that they didn’t ask for.
AEC firms don’t run on 2% margins because they’re slow to embrace innovation. They run on 2% margins because the value chain is structured that way. Risk tolerance is a luxury their P&Ls don’t allow.
That’s the room AI is walking into. Not a room of skeptics, but a room of operators who can’t afford to be wrong.
Our 2026 AEC Technology Outlook found 27% of firms using AI in their workflows. I’ve watched people in tech read the other 73% as resistance. They’re reading it wrong. Those firms aren’t refusing AI; they’re waiting for AI that doesn’t ask them to relearn their jobs.
The beta made three conditions plain — not because we set out to prove a thesis, but because the firms who got the most out of Bluebeam Max kept showing us the same things. What follows isn’t my opinion, but what more than 2,000 beta users did with the product before we ever launched it.
1. It has to live where the work already lives
The top barrier to digital adoption in our survey wasn’t cost. It was integration complexity — ahead of budget, leadership buy-in and skills gaps.
That tracks with every customer meeting I sit in. Nobody is asking for another tool; they’re asking for the tools they already have to do more.
Martin-Harris Construction is a Las Vegas construction management firm that ENR ranks among the top contractors in the West. Their projects span multifamily, industrial, educational, health care and high-rise work — which means their preconstruction team lives inside revision cycles.
During the Bluebeam Max beta, they used Smart Overlay and Smart Review to compare drawing revisions for pricing and impact analysis across both design and construction. Not a new workflow. Their existing workflow — revision comparison — with the manual hunting removed.
James Miranda, their director of preconstruction, described the value as the ability to “instantly see what’s changed and where to focus.”
That phrase is worth sitting with. He didn’t say the AI replaced anyone’s judgment. He said it pointed the judgment at the right sheet.
2. It has to earn trust before it asks for data
This one is harder, because trust isn’t a feature. You can’t ship it. You can only behave your way toward it. The firms not yet using AI keep naming the same concerns: who owns the data, how compliance works, how to make sure AI isn’t used irresponsibly. These aren’t paranoid concerns; they are the correct concerns.
The legal frameworks for AI accountability haven’t caught up with the products, and a contractor who carries liability at the end of a project is right to ask who carries it alongside them.
I have an obvious stake in saying this. Bluebeam benefits if contractors trust software with their data. But the trust problem is real whether or not I’m the one pointing at it.
The way through it isn’t promises but architecture. Bluebeam Max is built so the heavy lifting happens inside Bluebeam, and only the cleaned, restricted information moves beyond it.
A customer in our beta program told us they trust Bluebeam more than they trust the underlying language model — not because the model is bad, but because they know what we do with their work, and they’re not yet sure what anyone else does. That’s not a comment I would have predicted. It’s also the only kind of trust that compounds.
Steve McElwee, director of construction at Foresight Custom Homes, put the practical version of this well: having AI review the drawings alongside your own process — not instead of it — is what lets you bid with confidence that nothing slipped through. He called it “a gamechanger for preconstruction.”
3. It has to give time back — not take attention
I have a rule, internally, for any AI feature we ship: if it doesn’t return time, it shouldn’t exist. Not improve productivity. Return time. Hours, off the calendar, that the user can spend somewhere else.
KPFF’s Los Angeles civil office — a civil and structural engineering firm — used the beta to streamline day-to-day design tasks. Not a moonshot so much as the daily stuff: the repetitive markups, the drawing prep, the manual effort that fills the gaps between actual engineering.
That’s the work Magic Markups kills — duplicate a markup across material types, convert between measurement types, offset parallel lines CAD-style, all without leaving Revu. Stitching combines long corridor drawings into a single continuous view, which matters enormously if you’re a civil engineer and your project is 4 miles long.
And Bluebeam Max’s integration with Anthropic’s Claude — via MCP, the Model Context Protocol — lets a Revu user write natural-language prompts for things that used to require either deep software expertise or an afternoon of clicking. One beta user reported saving four to six hours a month on bookmarking and page labeling alone.
Our head of product, Luke Prescott, describes what we kept seeing in the beta: customers taking the tooling they already had and suddenly getting much more out of it. That’s the whole philosophy in one observation. We didn’t hand 2,000 people a new job. We handed them their old job, shorter.
The unglamorous version
I won’t tell you Bluebeam Max is going to disrupt construction. The word disrupt has done enough damage. What I’ll say is that the early adopters feedback we kept hearing wasn’t “this is interesting.” It was “this is what I needed last month.”
The temptation, always, is to lead with the vision. I believe AI will eventually run entire operations. I also believe the way you get there is by removing 20 minutes from one estimator’s day. Then another 20. Then another. Vision without those baby steps is just a press release.
The firms in our beta program weren’t customers, exactly. They were co-architects — the ones telling us what works and what doesn’t, in production, on real projects. That feedback loop is the actual product. The software follows.
I have immense gratitude for everyone in this industry who has trusted us to carry a little less for them. We do not take that lightly.
Join Our Newsletter Today
