Best of LinkedIn: Digital Construction CW 08/ 09

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 eight nine.

00:00:08: Frenness is a B to be market research company that supports enterprises across the construction industry with the market customer and competitive insights they need to navigate dynamic markets and drive customer centric product development.

00:00:21: That is a great overview of who was bringing us this information.

00:00:25: Yeah, and welcome everyone to This Deep Dive!

00:00:28: Today we are focusing on very specific two-week window that's calendar weeks eight and nine of twenty twenty

00:00:36: six And looking at the stack of research you brought today I have Really different this time around.

00:00:43: It really does.

00:00:44: yeah, usually when we look at these digest periods We are basically drowning in announcements about you know shiny new toys

00:00:51: right?

00:00:51: it's always looked what this magic tech can do

00:00:53: exactly But the conversation over these past two weeks feels surprisingly sober.

00:00:58: It's a lot less about the Magic and much more about well Why is this actually hard to implement?

00:01:03: that Is the perfect way to frame it.

00:01:04: I think The industry has collectively hit That moment where the hype fades And the real work begins.

00:01:11: We aren't really asking if the tech exists anymore.

00:01:14: We know it exists!

00:01:15: Right, now we're actually

00:01:25: looking at what is working in digital construction from AI on-the-job site to the reality of robotics and a major shift how we handle BIM and data.

00:01:35: So let's jump right into that first point

00:01:39: Because for the last year, the narrative has been that AI will just solve everything

00:01:42: right?

00:01:42: Oh absolutely.

00:01:43: But there was this post by Adolia Zhao That I think really stops that hype train in its tracks.

00:01:49: She calls it The Eleven P meter test.

00:01:51: Eleven p-meter tests!

00:01:52: i love them.

00:01:52: It is so visceral isn't it?

00:01:54: Zao's point Is that it is incredibly easy to demo an ai tool on a clean and perfect data set In conference room at ten in the morning Yeah

00:02:02: where Everything is structured perfectly

00:02:03: Exactly.

00:02:04: but that does not when A general contractor actually needs help.

00:02:07: No, a GC needs help when things have completely gone off the rails.

00:02:11: Right.

00:02:12: so The Eleven PEBR test is that scenario where an operations manager Is sitting in a trailer late at night digging through A messy PDF plan set trying to figure out why the concrete poor scheduled for tomorrow morning Does not match the architectural drawings.

00:02:28: and usually those plansets are just chaotic.

00:02:29: I mean they Have manual markups different versions red lines.

00:02:32: It's Not a neat database

00:02:34: precisely.

00:02:35: And Zhao argues That Most AI tools just choke on that reality.

00:02:39: They can't handle the noise, she frames it not as an adoption problem

00:02:42: Which is what tech companies always complain about by the way.

00:02:44: Totally they say construction too slow to adopt But she says its a signal extraction problem.

00:02:50: If your AI cannot extract truth from a messy real world document at eleven PEB It's just a toy.

00:02:58: So the barrier isn't that construction people are luddites.

00:03:00: Its that the tool aren't robust enough for chaos of job site.

00:03:04: That is such a crucial distinction for you listening to understand.

00:03:07: It really is and it pairs perfectly with what Tyler Campbell was saying during these same two weeks.

00:03:12: He warns that if you just throw AI tools at a chaotic process, You don't get efficiency...you just get chaos faster.

00:03:19: Chaos at scale?

00:03:21: he used the phrase mapping The Process which sounds a bit corporate.

00:03:24: but What does that look like practically on site?

00:03:27: It means identifying the bottleneck before you buy the software.

00:03:31: Campbell's argument is all about capacity versus clarity.

00:03:35: Adding more tools gives you more capacity to generate data, sure but it doesn't give you clarity on what that data actually means.

00:03:42: Right?

00:03:42: so if your RFI process has broken using AI to generate RFIs instantly just floods the architect with thousands of questions they can't answer.

00:03:51: You've just moved traffic jam, you haven't cleared

00:03:53: it.".

00:03:54: Exactly!

00:03:54: Have to know where apply pressure.

00:03:56: But let's play devil's advocate for a second.

00:03:58: There are plenty people out there who knows their process is flawed but they're terrified because digital transformation sounds like five year multi-million dollar headache.

00:04:07: And

00:04:07: that exactly why Hans Tepsel post was so refreshing.

00:04:11: He basically said to stop overthinking it.

00:04:13: Yeah, he shared a workflow that wasn't some massive enterprise-grade implementation involving dozens of consultants and committees... You just

00:04:21: walked the site right?

00:04:22: He recorded voice notes on his phone describing what he saw And used consumer AI tool To transcribe into structured cost tracking spreadsheet.

00:04:31: Just VoiceToStructuredData.

00:04:34: That is key there.

00:04:36: His insight was fascinating.

00:04:38: The tool teaches you how use.

00:04:40: You don't need to be a pumped engineer or learn how to code.

00:04:43: You just talked it like a colleague?

00:04:44: Exactly, if the output is wrong you say put the concrete cost in column D and it fixes it!

00:04:49: It lowers barrier entry to almost zero.

00:04:51: That sounds incredibly accessible.

00:04:54: But on other end of this spectrum we have Fernando Mayterena.

00:04:57: He wasn't doing simple spreadsheets.

00:04:59: he was something that frankly sounds magic.

00:05:01: It does sound like magic, but he showed the receipts.

00:05:04: He shared a workflow where he took a standard set of residential PDF drawings just two D lines on a page and converted them into a fully buildable Revit model.

00:05:13: for those who don't live in BIM software every day clarify The difference there.

00:05:17: why is the word buildable?

00:05:19: The key takeaway?

00:05:20: great question often AI can create amassing model.

00:05:25: That is basically a three-D shape that looks like a house but it's completely hollow.

00:05:28: It's a video game asset, right?

00:05:30: Metering his workflow generated over a thousand specific elements the framing studs The window schedules the drywall quantities Real actionable data you can actually use for estimation.

00:05:41: and

00:05:41: what was the time frame on that?

00:05:42: under ten minutes.

00:05:43: Wow

00:05:45: That is days of work for a human modeler.

00:05:47: Maybe even a full week

00:05:48: easily and this really highlights that the input phase Of construction getting the data into the system is compressing rapidly, but And there's very big butt here Tom Whitaker through a massive bucket of cold water on this whole excitement.

00:06:02: Yes storm chasers

00:06:04: right Whitaker is warning the industry about opportunistic startups.

00:06:08: These are software companies that see all the venture capital flowing into construction tech, so they slap a Construction AI label on their generic

00:06:15: product.

00:06:15: But they have no construction DNA?

00:06:17: Exactly!

00:06:18: They don't know the difference between a screed and a scaffold.

00:06:20: Yeah...and Whitaker's advice to the buyer is crucial here Do not bet your operations on company might vanish in eighteen months when the venture capitol dries up.

00:06:30: You

00:06:30: need partners who understand why a crew does things a certain way.

00:06:34: If can speak language of trade Show them door.

00:06:38: That really seems to be the dividing line in this whole AI discussion.

00:06:42: Does The Tool respect the reality of work?

00:06:46: Let's pivot where it physically happens.

00:06:49: This is our second theme today, Robotics and Automation!

00:06:52: And looking at funding news from weeks eight-nine... It looks like money will move from science projects into heavy industry.

00:07:00: The big headline grabber here was Lake Fisher Post about bedrock robotics.

00:07:05: They just raised a series B round

00:07:07: Two hundred and seventy million dollars.

00:07:09: That is not a let's see if this works amount of money.

00:07:11: that as we are scaling money

00:07:13: It is massive, but what's interesting?

00:07:15: Is their strategy?

00:07:16: they aren't trying to sell you a brand new futuristic autonomous bulldozer.

00:07:21: They are retrofitting

00:07:22: upgrading existing fleets.

00:07:23: yes And think about the economics of that.

00:07:26: A heavy earth mover costs hundreds of thousands, maybe millions of dollars.

00:07:31: a contractor is not going to scrap their entire fleet just to get autonomy.

00:07:35: Dead Rock is basically saying keep your machine we'll Just give it a brain.

00:07:37: exactly.

00:07:38: It signals that investors genuinely believe We are ready for driverless heavy equipment on real sites Not just in closed testing grounds.

00:07:47: So that's heavy civil work.

00:07:49: Yeah But we also saw updates on vertical construction.

00:07:52: Gothen Laying was posting about the Hadrian-X, which is The Brick Laying Robot.

00:07:56: We've seen videos of this thing for years but what's a new takeaway here?

00:08:00: The take away is the shift in human role.

00:08:03: Laying points out that we aren't replacing the Mason—we are elevating them!

00:08:07: When you use a Hadrian X... ...the Human isn't manually lifting up forty pound block every thirty seconds.

00:08:12: They're managing logistics checking quality overseeing whole system.

00:08:17: It moves from manual labor to technical supervision which extends their career, frankly.

00:08:23: It's much safer.

00:08:24: but we do have to be realistic here.

00:08:27: Wasim Rehman shared an analysis of some Chinese construction robots that are designed to install ceiling panels and this goes right back to our messy site thing We talked about with AI.

00:08:36: Let me guess they work great in a pristine lab.

00:08:39: They work beautifully in a lab.

00:08:41: But Raymond pointed out that on a real site, the floor is covered in dust cut wires maybe a discarded lunchbox.

00:08:48: there are other trades walking around with ladders

00:08:50: and The robot just gets confused or

00:08:51: it get stuck Or worse It becomes a safety hazard.

00:08:55: It turns out that navigating A dirty floor Is actually a harder engineering challenge than screwing In the ceiling panel itself.

00:09:01: So is this solution to Just clean the Floor?

00:09:08: Tobias Claus highlighted the BowBot, which automates fastening by drilling and marking holes based on BIM data.

00:09:14: That seems to handle real-world environment a bit better.

00:09:17: but Shadi Khuzam writing from Dubai argues we need much bigger shift.

00:09:22: he proposed the hybrid model.

00:09:24: He says that We Need To Stop Looking For One Robot That Builds The Entire House.

00:09:29: What's Right Golden?

00:09:30: Systems Integration You use three d printing for the vertical walls where it makes sense.

00:09:35: Use precast concrete slabs, because that's cheap and fast.

00:09:40: Use prefabricated pods of the bathrooms.

00:09:42: you used right method.

00:09:45: instead trying to force a robot to lay bricks.

00:09:48: maybe just shouldn't be using bricks in first place.

00:09:50: if you want automation exactly

00:09:52: change them at suit the automation.

00:09:54: don't try to automate old manual methods.

00:09:56: is systems integration challenge not robotics?

00:10:01: Which brings us to the backbone of all this.

00:10:03: If you're gonna have precast slabs talking to three D printers and AI checking the schedule, You need data.

00:10:10: And that leads us to our third theme with his deep dive BIM digital twins and standards.

00:10:16: This is the unsexy plumbing that makes everything else possible, and Santosh Kumar Boda had a fantastic analogy to explain where we are heading.

00:10:24: he said to think of bim building information modeling as The Skeleton!

00:10:28: The geometry...the beams..the column

00:10:31: Right but a skeleton can't feel anything.

00:10:33: so He introduces IOT sensors in data As the nervous system And the goal Is To get this body ready for what he calls day fifty four.

00:10:41: Why day fifty-four?

00:10:42: It's

00:10:42: symbolic.

00:10:43: Day one is design, day fifty is construction.

00:10:46: Day fifty four is when the contractor has packed up and left The keys are handed over And the facility manager Is alone in building.

00:10:52: Ah...the operations phase.

00:10:53: Yes

00:10:54: If a pump starts vibrating itself to death on day fifty fourth A digital twin needs know.

00:10:59: A bonus point that an alert shouldn't just be a line of text In a database nobody ever checks.

00:11:04: it should be a flashing red box in room.

00:11:05: three oh two Of the three d model.

00:11:09: But building that nervous system sounds incredibly complex.

00:11:13: I mean, you have fifty different software vendors involved in a single building.

00:11:17: That is exactly the problem Florian Hummer addressed.

00:11:20: He warned against all-in-one fantasy.

00:11:23: You know The sales pitch where a vendor promises their platform does everything

00:11:27: It ever does.

00:11:28: Never Humor outlines a four-layer stack that you actually need to build.

00:11:31: You need the platform layer for holding the data, you need a physics layer for simulation...you know will this wall actually hold the weight?

00:11:39: ...You need a logic layer for optimization and an ROI layer for financial diagnosis.

00:11:44: You're

00:11:45: building a tech stack!

00:11:46: You are not buying a single off-the-shelf solution.

00:11:48: Correct.

00:11:49: And because you were building a stack with different vendors The standards matter more than ever.

00:11:55: This is where Nikola Jovic dropped a pretty significant update regarding ISO-Rx.

00:11:59: I

00:12:01: know ISO standards can be dry, but Jovic flagged the change coming in March.

00:12:05: that sounds incredibly important for anyone listening who works on international projects.

00:12:10: It's vital!

00:12:12: The shift is subtle and profound... ...the standard is moving language away from BIM which implies modeling geometry to information management.

00:12:22: What is practical difference there?

00:12:24: It changes the entire mindset from delivering a file to managing an asset.

00:12:30: The Asset Information Model becomes permanent requirements, not something you hand over on USB drive at end of job and just forget about it's living database that stays with building for fifty years.

00:12:40: So data has survive longer than construction company does?

00:12:43: Exactly!

00:12:44: And manage massive amount.

00:12:46: during design phase.

00:12:47: we are seeing AI pop up again.

00:12:49: Christochon Velibut shared a workflow using AI agents in Revit, and notice I said Agents not Assistance.

00:12:57: That

00:12:57: distinction is becoming really important.

00:12:58: an assistant waits for you to ask the question.

00:13:01: An agent acts on its own right?

00:13:03: Yes it acts autonomously based upon your goal.

00:13:05: You give it!

00:13:07: Velibuch described as setup with four agents running simultaneously.

00:13:11: One is checking wall parameters.

00:13:13: Another is validating room names, another is checking scope boxes.

00:13:17: It's literally like having an army of interns who are constantly proofreading your work in real time.

00:13:22: That

00:13:22: sounds incredibly efficient But it also sounds a bit dangerous if the underlying instructions or bad

00:13:28: Which leads perfectly to Mario Enrique's Rebello point about The BEP-The BIM Execution Plan.

00:13:35: Oh!

00:13:35: The document everyone writes and absolutely nobody reads?

00:13:37: Usually

00:13:38: yes.

00:13:39: that Is what he calls THE BAD BEP A static document created just to satisfy a contract requirement.

00:13:45: Robelo contrasts this with the good BEP, which is dynamic.

00:13:48: it actually defines The workflow.

00:13:50: his point Is if your plan is bad?

00:13:53: Your data will be Bad.

00:13:54: and If you're Data as Bad You're AI agents are Just Gonna Make Mistakes A lot Faster.

00:13:58: It Always Comes Back To the Plane.

00:13:59: You Cannot Automate A Bad Strategy?

00:14:01: No!

00:14:01: You Really Can't.

00:14:03: And That Is The Perfect Segway To Our Final Theme Strategy Leadership and Regional Momentum Because you can have all the ISO standards and the two hundred seventy million dollar robots.

00:14:14: Yeah, but if people on site don't buy in.

00:14:16: The whole thing fails

00:14:18: Absolutely.

00:14:19: And if you want to see where the culture is shifting?

00:14:21: The fastest right now You have to look at the Middle East.

00:14:24: Marcus Goddrowy and Adele Miran both shared really deep insights On this.

00:14:28: they referred to Saudi Arabia and the UAE as proving grounds.

00:14:31: yes And it's not just because they have massive amounts of capital, Gadar we noted that the entire system is aligned.

00:14:38: You have the regulator setting the rules you have the capitol flowing and you have contractors all moving in sync to industrialize construction

00:14:46: unlike in other regions where the regulator might be fighting the contractor, or the capital is highly risk-averse.

00:14:51: Exactly!

00:14:51: They are using these gigaprojects to test things at a scale and speed that you simply do not see in Europe nor North America.

00:14:57: right now they're forcing evolution through sheer brute force an aligned strategy But

00:15:03: globally whether you were in Riyadh London New York there was major talent shortage And Oliver Pohl had a really interesting observation about how tech companies are responding to this.

00:15:15: I was very happy to see his post, Paul notes that tech companies were finally pivoting who they hire.

00:15:20: for years They just hired saw salespeople people Who could sell software but had never set foot on a muddy job site.

00:15:27: now?

00:15:28: There hiring for domain expertise former superintendents former project managers

00:15:33: because the superintendent can smell nonsense a mile away.

00:15:36: That's it.

00:15:37: As Paul says, a superintendent doesn't care about your tech stack or open API.

00:15:41: They care if the tool works.

00:15:43: when there is no internet connection in the basement they care of it makes their crew safer.

00:15:47: You need people selling the Tech who have actually worn boots and felt pain.

00:15:51: But even once you buy the Tech from the right person... ...you need leadership to ACTUALLY make it stick.

00:15:56: Tony Obujwaad had pretty harsh truth where digital transformation fails.

00:16:02: It was fantastic Leadership lesson.

00:16:04: He said Digital transformation doesn't die in the boardroom.

00:16:07: It dies in the field.

00:16:08: it dies The exact moment a site manager opens WhatsApp instead of the platform.

00:16:13: you just spent millions Of dollars on.

00:16:15: that is such a vivid image, but what's up versus platform battle?

00:16:19: and let's be honest What's app usually wins?

00:16:20: because?

00:16:21: Because it's so easy

00:16:22: it winds because it Is incredibly low friction.

00:16:25: Jawad's takeaway is that leaders cannot just mandate software from the head office.

00:16:29: You have to be there, you have to show up on site look The Site Manager in the eye and show them exactly why the platform helps them.

00:16:36: if you aren't willing To get your boots dirty Your digital strategy Is Just a PDF.

00:16:41: So pulling this all together for you listening.

00:16:43: We've covered a lot of ground from these two weeks of insights.

00:16:46: We

00:17:06: are done with the wow phase.

00:17:08: we're past the hype cycle where we just share cool videos of robots doing backflips.

00:17:13: now, we are

00:17:13: firmly in

00:17:14: the how-phase.

00:17:15: How do you make AI resilient to dirty data?

00:17:19: How do we integrate robots with human crews without slowing them down?

00:17:24: How did build business models that actually support this transition?

00:17:27: It sounds a little less exciting than sci fi but it's much more important.

00:17:31: It's harder work than just watching a demo, but it is the work that actually changes.

00:18:01: Well, that is a wrap on this deep dive.

00:18:03: If you enjoyed this episode new episodes drop every two weeks.

00:18:06: also check out our other editions of Smart Manufacturing and Digital Power Tools.

00:18:10: Thanks so much for listening.

00:18:11: Thank You!

00:18:11: And don't forget to subscribe So you won't miss the next Deep Dive into the industry's best insights.

00:18:16: See ya next time.