Best of LinkedIn: Digital Construction CW 14/ 15

00:00:00: This episode is provided by Thomas Allgaier and Frennus based on the most relevant LinkedIn posts about digital construction in calendar weeks, fourteen and fifteen.

00:00:08: Frenness is a BDB market research company that supports industrial automation and ICT companies with market intelligence across the construction industry to prioritize segments identify high value accounts and validate use cases.

00:00:22: you can find more info in the description

00:00:24: right so let's dive into it.

00:00:26: welcome to our latest deep dive.

00:00:28: Today, we are extracting the absolute top digital construction trends that we've been seeing buzzing across LinkedIn.

00:00:35: Yeah

00:00:35: and for you listening We have gone through all of this noise And clustered these insights into a few really critical themes.

00:00:41: So no fluff today.

00:00:42: Just to focus.

00:00:43: look at what is actually working in Construction Tech right now

00:00:46: Exactly, because there is a ton of hype out there.

00:00:49: But we want to focus on the reality on the ground

00:00:51: which perfectly brings us To our first theme Which is how data foundations are turning into a massive competitive advantage?

00:00:57: Because There's this huge contradiction in The industry right now.

00:01:00: like everyone wants the shiny new AI tools but the Groundwork simply isn't their.

00:01:05: Yeah, it's a huge issue.

00:01:06: Fabio Loria was actually talking about this recently.

00:01:08: he pointed out that the AI in construction market is projected to hit something like fourteen point.

00:01:14: five billion dollars by twenty thirty-two.

00:01:16: Wow!

00:01:16: Fourteen point five million

00:01:17: right?

00:01:18: Huge number.

00:01:19: but then he hits you with The Sobering Reality which is that half of industry still relying on paper.

00:01:24: It's wild Half the Industry.

00:01:25: and Ann Berger had an observation ties into that.

00:01:28: She noted most companies are essentially running hundreds of scattered Excel sheets.

00:01:34: Oh,

00:01:34: the dreaded spreadsheets!

00:01:36: Right you've got an estimator with one version and a site manager for another.

00:01:39: And her point was that AI cannot do anything until this unstructured data is actually centralized.

00:01:45: Yeah...and thats exactly why AI initiatives stall out.

00:01:48: AJ Waters had really great take on it.

00:01:50: He said that AI doesn't fail because weak models.

00:01:53: It fails due to garbage data

00:01:55: Garbage and garbage out.

00:01:56: Exactly.

00:01:58: We're talking about siloed ERPs an unreadable PDS that nobody can parse.

00:02:03: So it's not a technology problem we're facing, It is data governance problems.

00:02:07: To put in perspective for you listening...it like trying to run high performance Ferrari engine on unrefined crude oil

00:02:14: That great way of putting it Like

00:02:15: before going by the Ferrari.

00:02:17: You absolutely need build refinery.

00:02:21: If Data Consolidation is step one Where are companies actually applying this refined data first?

00:02:29: Well, the smartest firms aren't waiting until a project is already a mess.

00:02:33: They're applying that clean data at the very beginning of life cycle.

00:02:37: So our second theme today Is how pre-construction moving to the strategic center Of everything

00:02:43: Right because pre construction isn't just about throwing together an estimate anymore.

00:02:47: No not at all.

00:02:47: It's essentially becoming The control layer for entire project.

00:02:51: Lisa Mofapatra brought this up.

00:02:52: actually She said pre construction has become AI native decision system

00:02:57: an AI-native decision system.

00:02:59: That's a strong phrase!

00:03:00: It is, and she backed it up by pointing out that things like prefab... ...and modular construction.

00:03:07: uh we usually call those trends right but she argues they aren't trends at all.

00:03:10: They are actually the outcome of having absolute certainty during preconstruction.

00:03:14: Oh

00:03:14: that makes total sense because if your data is perfectly clean before you build You can confidently manufacture a whole wall offsite.

00:03:22: Exactly The certainty is what enables the method And tech platforms are totally leaning into this.

00:03:28: Aaron Kramer posted recently about Trimble acquiring document crunch.

00:03:32: Oh yeah, I saw that they've been used on over ten thousand projects or something

00:03:36: Over ten thousand?

00:03:37: Yeah and the takeaway there is these tech platforms aren't just sitting around waiting for general AI to mature.

00:03:43: They're actively buying up deep workflow specific intelligence To bake it right in the DNA of their suites

00:03:51: Which is incredible to see.

00:03:52: It's not just theory.

00:03:54: Yeah.

00:03:54: Ashley Huberney shared some numbers that honestly blew my mind.

00:03:57: Oh, about Forman AI?

00:03:58: Yeah.

00:03:59: Forman Ai in just its first week processed ninety-four million dollars In real construction budgets and it handled two hundred fifty six million Dollars in project value...in one week.

00:04:11: so It just proves that ai is actively handling the heavy lifting in The office layer right now.

00:04:17: That Is a staggering volume of work!

00:04:20: But it begs a huge question, right?

00:04:22: If AI is taking over the office layer and handling literally millions in budgets how should a construction firm even budget for this technology?

00:04:31: because clearly isn't just an IT expense anymore.

00:04:34: And that question leads perfectly into our third theme which is AI becoming core construction infrastructure.

00:04:40: and of course there risks involved with that Because There Is A Total Paradigm Shift Happening In How We View AI.

00:04:46: Okay what do you mean by paradigm shift?

00:04:48: Well, we're moving away from treating it as a software subscription and starting to look at it as virtual workforce.

00:04:54: Virtual Workforce?

00:04:55: I like that!

00:04:55: Yeah, Christopher Langeza framed it brilliantly.

00:04:57: he said AI is a labor cost not a software costs.

00:05:01: Oh thats interesting.

00:05:02: so ships budgets entirely.

00:05:03: Exactly think about it A superintendent out on the site.

00:05:06: They don't call IT department to approve a five hundred thousand dollar dozer rental.

00:05:12: Right they just assigned the cost of job

00:05:14: Exactly.

00:05:15: So, Linjaze's point is that if AI agents are reviewing submittals or drafting RFIs... ...that should be treated the exact same way as project labor.

00:05:25: That makes so much sense But it also changes the dynamic of what the AIs allowed to do.

00:05:30: Nate Fuller was writing about this right?

00:05:31: The whole concept of agentic AI?

00:05:34: Yes!

00:05:34: Agentic AI like Claude Cowork Fuller pointed out that these agents can actually access local project files.

00:05:41: They could read, write and even delete data on your drives!

00:05:43: Wait wait.

00:05:44: if an AI agent can actively access or alter a local project file the risk profile completely changes.

00:05:52: I mean are we just trusting it to make commercial decisions based whatever it finds?

00:05:56: That is exact fear.

00:05:58: you should be alarmed because Simon Rhodes brought up psychological risks here.

00:06:04: He talked about something called delusional spiraling.

00:06:06: Delusional spiral, what is that?

00:06:08: It's

00:06:08: based on some recent MIT research.

00:06:10: basically it shows the AI tends to tell users What they want.

00:06:13: here wants be helpful.

00:06:15: oh That sounds dangerous in construction.

00:06:17: very because if an AI just Validates and overly optimistic margin estimate instead of you know actually challenging it its not managing commercial risk.

00:06:25: Its hiding

00:06:26: right.

00:06:26: so I say hey AI totally pour this concrete in two days.

00:06:30: The area just goes, yep looks great to

00:06:32: me.".

00:06:32: Exactly and then you spiral into this delusion that your schedule is fine.

00:06:37: Rhodes made the point that the real winners of space will be companies who use AI to surface uncomfortable truths not validate bad estimates.

00:06:46: That's such a critical point!

00:06:48: You need an AI pushback.

00:06:50: but look we've talked about the digital side office layer budgets.

00:06:54: Let's transition to the physical job site, because how are we actually connecting these hyper-intelligent digital plans... ...to The Mud and The Dirt?

00:07:02: That is theme number four.

00:07:03: Bridging into the Job Site with Reality Capture & Robotics And Chaitanya NK had a really profound realization about this.

00:07:10: Yeah what did he say?

00:07:11: He said models describe intent.

00:07:13: reality describes truth.

00:07:15: Oh man!

00:07:16: Models Describe Intent.

00:07:17: Reality Describes Truth.

00:07:18: that Is Powerful Right.

00:07:20: For so long,

00:07:21: we've tried to force reality to match the BIM model.

00:07:24: but he argues that you have to actually measure what physically exists

00:07:28: And Florian Humeur echoed this too right?

00:07:30: He laid out a whole four step architecture for digital twin!

00:07:33: He did and his main point was if it doesn't have live data feeding into it is basically just a video game asset.

00:07:40: Yeah It looks pretty But its useless for actual building.

00:07:45: Which brings us to the physical robotic side.

00:07:47: Tobias Claus had a post recently that I got really excited about.

00:07:50: he was talking About dusty robotics.

00:07:52: Oh,

00:07:52: I saw that they're doing some amazing things.

00:07:54: They literally print BIM layouts Directly onto the job site floors like the robot just drives around and prints The digital model under the concrete.

00:08:03: it's brilliant.

00:08:04: It completely removes the human error from the physical layout process And Varad Mughod actually backed this up with a Zekwa Ventures report.

00:08:12: He showed the construction.

00:08:13: robotics are seeing real scale right now in specific areas like

00:08:17: where.

00:08:17: just lay out layout definitely but also solar installations rebar tying and site capture, get this they're delivering thirty to fifty percent labor savings.

00:08:30: thirty two fifty percent.

00:08:31: that's massive for an industry.

00:08:33: They can't find enough skilled workers.

00:08:36: To clarify for the listener, it's not like The Terminator is out there replacing the whole crew.

00:08:40: No definitely no!

00:08:41: It's more of a highly precise power tool that just amplifies the skilled trades on-site.

00:08:48: I completely agree with Power Tool analogy but i do have to issue warning here which leads into our final theme Field Fit and Human Element.

00:08:55: because honestly doesn't matter how great tech is if humans absolutely refuse.

00:09:00: pick it up.

00:09:00: So true The adoption bottleneck, Zolkanane Malik shared a brutal truth about this.

00:09:05: He said most construction software dies on Monday morning.

00:09:08: It's such great quote because it is so true.

00:09:11: Yeah like if he takes too many clicks out in the field then crew will just say forget and go back to sending text messages to site manager.

00:09:17: Exactly!

00:09:18: Stephanie Brown shared some data that totally supports That.

00:09:21: She noted only twelve point two percent of businesses are actually using AI right now.

00:09:26: Just twelve-point two percent?

00:09:28: Yep which just proves that the field workers, the men and women actually pouring the concrete they still hold the ultimate veto power over what technology actually

00:09:37: sticks.

00:09:38: Which is why Jacob Schaevel's concept is so important.

00:09:40: he brought up this idea of ambient data collection

00:09:43: Ambient Data Collection?

00:09:44: How does that

00:09:45: work?

00:09:45: Well basically you use AI to passively trawl through text threads in emails So the field team doesn't have to input anything manually.

00:09:53: They just chat like they normally do and the AI extracts The schedule updates or the delays right from their normal communication.

00:10:00: that is genius.

00:10:01: It completely removes the friction of adopting a new software interface.

00:10:04: Exactly, you don't know if it train anyone but I think Heather Adkins in Brett King Really summed up this whole human element perfectly.

00:10:11: What did they focus on?

00:10:12: They just reminded everyone that B-to-B buyers at the end of the day are still humans.

00:10:17: Construction is always going to be a business that's built on shared experience, tough jobs and deep

00:10:23: trust.

00:10:23: Oh hundred percent!

00:10:25: A robot can't shake your hand and promise to make it deadline?

00:10:27: Exactly so for you.

00:10:29: listening as we wrap up this deep dive I want leave with one final kind of mind-bending thought.

00:10:33: Let us hear

00:10:34: This comes from LuY's research on the DreamHouse benchmark.

00:10:37: Despite all these hype about AI taking over They tested best frontier AI models on generating structurally valid Timber frame houses like houses that actually pass real engineering codes.

00:10:51: Okay,

00:10:51: and how did the AI do?

00:10:53: It scored a dismal seven point one percent seven

00:10:56: point One percent.

00:10:56: now

00:10:57: it completely failed.

00:10:58: AI still lacks basic physical reasoning.

00:11:01: so ponder this.

00:11:02: AI can crunch your budget.

00:11:04: They could read you're hundred-page contracts And I can track your data.

00:11:07: The physical laws of gravity in structural integrity.

00:11:10: those still belong entirely to the human engineer.

00:11:13: That is the perfect reality check.

00:11:15: to end on

00:11:15: If you enjoyed this episode, new episodes drop every two weeks.

00:11:19: Also check out our other editions on smart manufacturing and digital power tools.

00:11:23: Thank You for listening!

00:11:24: And don't forget to subscribe.