Best of LinkedIn: Digital Construction CW 06/ 07

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 six and seven.

00:00:09: Frenness is a B-to-B 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: Welcome To The Deep Dive.

00:00:23: It's great to be back you know.

00:00:25: Looking at these last couple of weeks calendar week six and seven it feels like something shifted.

00:00:29: I

00:00:29: saw that too.

00:00:30: It's there's a different tone.

00:00:32: Yeah, it's like the industry has a collective hangover from the whole tech hype phase And is now I don't know getting serious about how to actually make things work.

00:00:40: That's a good way to put it.

00:00:41: for years It's been look at this cool new tool.

00:00:44: But now the conversation from folks like Roman H or dr Ben Jowett?

00:00:48: It's all about operations the messy business of fitting tech into a real-world process.

00:00:56: We're finally moving from The What to The How, and honestly it's about time.

00:01:00: we've got a lot to get through today!

00:01:02: We are going hit with the reality of AI.

00:01:03: something called Vibe Coding

00:01:05: which is fascinating term will also get in digital twins.

00:01:09: why your BIM model?

00:01:10: probably not twin?

00:01:12: And have talk physical build there this story data center.

00:01:17: A ten centimeter error that'll make your wallet hurt just hearing it.

00:01:20: Oh,

00:01:21: That one was painful to read.

00:01:22: It really was yeah.

00:01:23: So let's uh Let's start with the brain of the operation AI and data.

00:01:28: Let's do it okay?

00:01:29: so vibe coding.

00:01:31: Nate Fuller and Humza Shonbari were discussing this.

00:01:34: is This Just?

00:01:36: I don't know a new buzzword or Is there something real here?

00:01:38: Its surprisingly Real.

00:01:39: And its basically using natural language AI To build small internal tools in like minutes.

00:01:45: You don't write code, you just describe the vibe of what Exactly.

00:02:04: And project is over.

00:02:06: by then with vibe coding that engineer builds a prototype on their lunch break.

00:02:10: It's about empowering the people in the field to solve their own problems quickly.

00:02:15: I see the appeal.

00:02:15: for sure Yeah, but uh That sounds a little terrifying from a safety and governance perspective?

00:02:21: That's

00:02:21: the catch and they were very clear about this.

00:02:23: This is for prototyping its citizen development.

00:02:26: You absolutely need a human in the loop to check the work.

00:02:30: You're not running a nuclear reactor with a vibe coded script,

00:02:33: right?

00:02:33: It's a tool for iteration.

00:02:35: Not for final deployment without oversight Mm-hmm

00:02:38: But even with the best tools they need fuel and that fuel is data Which is where Roman H kind of poured some cold water on the AI height.

00:02:48: I'd

00:02:48: call it a reality check

00:02:49: A Reality Check, yeah His point is that AI Is basically useless for a lot of firms right now because all their data isn't silos.

00:02:56: You have your schedule here Your costs over there Drawing somewhere else.

00:02:59: So The AI can't see the full picture.

00:03:00: It's blind...it Can't or cross-reference a scheduled delay with cost Overrun to give you any real insight.

00:03:06: Its' All just theory at That Point

00:03:08: And Dr Benjel would took this even further.

00:03:09: He says not even a technology problem.

00:03:12: Right, it's a business strategy problem.

00:03:14: you can't just bolt AI onto a broken operating model and expect magic garbage in garbage out.

00:03:19: this connects directly to what Matt Gauss said which I think was maybe the most profound insight of the week

00:03:25: The legal ecosystem idea.

00:03:26: Yeah

00:03:27: he said construction isn't organized as an operational production system It's a legal ecosystem.

00:03:33: That distinction is everything.

00:03:35: Okay Break that down for me.

00:03:36: What does it actually mean?

00:03:37: Well,

00:03:37: think about a car factory.

00:03:39: It's production system.

00:03:40: right.

00:03:40: every single decision is optimized For the flow of the car down-the line Right

00:03:45: for efficiency.

00:03:46: exactly.

00:03:46: now Think About A job site.

00:03:48: We organize it around contracts The GC the subs the architect.

00:03:53: its all about transferring risk and liability.

00:03:55: so we optimize.

00:03:56: four don't get sued not build efficiently.

00:03:58: precisely an AI thrives in a Production System where data flows freely.

00:04:03: In our legal ecosystem, data gets hoarded as a defense mechanism.

00:04:08: So to really use AI that's point is we might have to completely redesign the industry's operating system.

00:04:13: That's huge thought!

00:04:15: Not just buying new software.

00:04:17: Speaking of software let us talk about market.

00:04:20: Guido Machiochi brought up this idea of kill zone.

00:04:23: This is classic David vs Goliath story.

00:04:26: Guido's warning is Autodesk has massive datamote Decades of Rebit files, BIM-M.

00:04:34: So

00:04:34: if you're a startup trying to build a new AI tool You are basically starving for the data that Autodesk is feasting on.

00:04:41: That's The Killzone.

00:04:42: If you build something great they can copy it cut off your API access or just buy.

00:04:46: It's incredibly difficult to compete.

00:04:48: But

00:04:49: Guido didn't just leave it at that, he had a wildcard,

00:04:51: didn't they?

00:04:55: He did!

00:05:07: but you could build a thousand tiny tools that chip away at their business.

00:05:10: That's the idea, which brings us back to that citizen developer idea and I want to ground this with a real example because we've been very high level.

00:05:17: Belal Musa shared this great story

00:05:19: The Brick Counting Story.

00:05:20: Yes!

00:05:21: The brick counting stories...I

00:05:22: love this one.

00:05:23: It was

00:05:23: brilliant.

00:05:24: So he wanted a quote To get his house rendered.

00:05:26: The contractor tells Belal Go measure it himself so He can give him a price

00:05:31: Which is just- I mean You're asking the customer Do do work its insane.

00:05:35: Its

00:05:35: pure friction.

00:05:37: So instead, Bilal takes an image of his house from Google Maps feeds it to Gemini and asks it to estimate the surface area by literally counting the bricks.

00:05:47: And knowing their standard size

00:05:49: you work.

00:05:50: he measured it manually To check.

00:05:52: It was spot-on.

00:05:53: that is amazing.

00:05:54: but The real insight isn't just cool tech.

00:05:57: its about sales.

00:05:59: That contractor could have given a rough quote over the phone in thirty seconds looks like x square meters.

00:06:04: Here's a ballpark.

00:06:05: That's how AI makes you money today.

00:06:07: It removes friction from the business.

00:06:09: that's a perfect segue.

00:06:11: We've talked about the AI brain.

00:06:12: now Let's talk about the digital body The Digital Twin.

00:06:15: or as Florian humor would say your BIM model is not a digital twin.

00:06:19: He was pretty forceful about that.

00:06:21: I feel like Digital Twin gets slapped on any three-D model just to sound expensive.

00:06:25: What's the actual difference?

00:06:26: it all comes down to whether its alive, Or Not.

00:06:29: A BIM Model Is Static.

00:06:30: Its Design Intent Humor lays out a five-layer architecture.

00:06:34: You have data acquisition, that's IoT sensors then transmission Then modeling.

00:06:39: That's where BIM fits.

00:06:40: So BIM is just one layer out of five?

00:06:42: Just One!

00:06:43: Then you have integration often in a game engine.

00:06:46: But the most important layer The one everyone misses Is the fifth

00:06:50: one... ...the service layer.

00:06:52: This is where value is.

00:06:53: If your model doesn't do anything if it can't predict an equipment failure or optimize energy use It's just pretty.

00:06:59: three D dashboard A true twin provides a service back to you, it predicts.

00:07:03: Right!

00:07:03: Dr.

00:07:04: Amal Nath Elniger had similar framework with his maturity levels.

00:07:08: He argues that most of us are stuck in Levels one-to three just getting our data together.

00:07:12: The real money the real ROI is up at level seven and nine.

00:07:15: That's when building tells your own HVAC filter as clogged And its already ordered replacement.

00:07:20: We're long way from there.

00:07:22: But before they can talk we have know where this is.

00:07:25: Santosh Kumar Boda made a great point about shifting to geospatial intelligence.

00:07:30: Why is wear suddenly so important?

00:07:32: Because for so long we've been obsessed with the how, How To Build The Thing that's bimmed afield.

00:07:39: Santosh is saying We need to master the where because That's Where The Biggest Decisions Are Made

00:07:44: Like What For Example

00:07:45: Climate risk is a huge one.

00:07:46: You can design a perfect building, but if you put it in a future floodplain... ...you've failed.

00:07:51: or deciding where to put a retail entrance based on foot traffic data.

00:07:55: It's moving from model-to-execution To location-to decision

00:07:59: about zooming out from the building to its entire context.

00:08:02: exactly and speaking of Context let's talk about existing buildings.

00:08:07: Bertha Zuzo glue From new begun dropped a stat that really grounds this whole conversation

00:08:12: which was at

00:08:13: Eighty-five percent of buildings in Europe were built before the year two thousand.

00:08:16: Wow, so all this talk about smart new builds?

00:08:19: Yeah It's irrelevant to the vast majority of our building stock unless we can retrofit.

00:08:23: That's the market reality.

00:08:25: We have to digitize what's already there and that's why reality capture is so critical.

00:08:29: You can't trust a thirty-year old drawing.

00:08:31: it's wrong.

00:08:32: Mario Enrique for below was talking about this.

00:08:34: with Scanda BIM.

00:08:35: Its basically reverse engineering.

00:08:37: The Building its

00:08:37: All About Reducing Design Assumptions.

00:08:40: And there's some cool tech evolving here.

00:08:42: Christogen Vildic shared a tool called Cobot,

00:08:45: the Google Maps thing?

00:08:46: Yeah it uses AI to generate the surrounding environment for new project directly from Google Maps' three-D data.

00:08:53: so an architect doesn't have spend days manually modeling neighborhood.

00:08:56: It is

00:08:57: context on demand.

00:08:58: Pretty much!

00:08:59: It just dissolves barrier between design and real world.

00:09:02: Okay we had the AI The Twin The Context But eventually someone has actually build that.

00:09:09: Let's move to execution, prefab and the human element.

00:09:13: This is where The Rubber Meets The Road... ...and Where It All Goes Wrong!

00:09:16: Let's start with that horror story.

00:09:18: Samer Ilsaia's data center example?

00:09:20: This is such a classic case of what happens when you separate construction & technology.

00:09:25: Samers' point Is That A Data Center isn't a building it's a machine.

00:09:30: people walk inside.

00:09:31: I like that its great frame.

00:09:32: So as civil contractor builds the shell concrete, steel.

00:09:36: It looks perfect all within normal construction

00:09:38: tolerance.".

00:09:39: Then the tech team comes in to install the server cooling corridors and they're misaligned by ten centimeters.

00:09:47: Ten centimeters?

00:09:47: That's what four inches!

00:09:49: In a house you could probably fudge that...

00:09:51: In high density data center it is catastrophe.

00:09:53: You can't just fudge pressure cooling systems.

00:09:57: The connections don't line up Air flow doesn't work.

00:10:00: So was the outcome?

00:10:01: Four weeks of rework Breaking concrete moving mounts, total disaster all because the civil contract and the tech contract weren't treated as one integrated system.

00:10:11: That goes right back to Matt Gauss point about the legal ecosystem causing these failures.

00:10:15: It's exact same problem in practice.

00:10:17: So how do we fix execution?

00:10:19: The Wandao had some really smart insights on prefab this idea of push versus pull.

00:10:23: Yes This is straight out of lean manufacturing pushes them onto the site.

00:10:35: And the site is never ready for them?

00:10:36: Never, yeah.

00:10:37: so you have these expensive finished pods sitting in the mud getting damaged getting in the way.

00:10:43: Dohan said rework can hit twenty percent which

00:10:46: completely defeats the purpose of doing it in a factory.

00:10:48: right The solution is pull.

00:10:50: You have to stabilize the site schedule first using something like tax planning.

00:10:55: For those who don't know tact is basically a rhythm isn't

00:10:57: exactly.

00:10:58: It's German for beat or pulse.

00:11:01: You create a steady rhythm of work on site and then you pull the prefab modules from the factory so they arrive just in time to be installed, The site dictates the factories pace.

00:11:12: That's the other way around

00:11:13: exactly.

00:11:14: but that requires incredible management.

00:11:15: And that brings us To the silent killers Of projects Erdem ever mentioned.

00:11:19: one I think is so common Ignoring crew input In planning.

00:11:23: Oh

00:11:23: this One Is So Basic.

00:11:26: It's

00:11:26: The Classic Disconnect.

00:11:28: Someone in an office builds a perfect schedule on screen, but they never ask the foreman if it's actually possible to do that work sequence.

00:11:37: And

00:11:37: then when the schedule slips... The crew gets blamed for being unproductive.

00:11:41: Which leads directly

00:11:43: to burnout!

00:11:44: he put a number on this.

00:11:47: He developed a tool around the concept called Context Switching Tax.

00:11:53: Yes,

00:11:54: I loved it!

00:11:55: We all kind of know multitasking is a myth but he actually quantified that cost.

00:11:59: He did?

00:11:59: His app shows you productivity when you split people across too many projects.

00:12:04: A PM for four project isn't giving twenty-five percent to each.

00:12:07: they're losing huge amounts of time just mentally reloading information for one.

00:12:12: He calls it Burnout ROI.

00:12:14: It puts a dollar amount on it.

00:12:15: It turns the complaint, I'm overwhelmed into a business case.

00:12:18: this staffing model is costing us fifty thousand dollars a

00:12:21: month.".

00:12:21: That's how you get management to pay attention.

00:12:23: and speaking of roles Alistair Lewis made the point that the BIM manager title might be dead

00:12:29: or at least evolving.

00:12:30: he suggests The new title Is digital design leader And the change in words is important.

00:12:36: A BIM Manager manages a tech stack.

00:12:38: A Digital Design Leader Manages a Tech Web The flow of data, the systems.

00:12:43: The business outcomes.

00:12:44: So it's

00:12:44: moving from technician to strategy

00:12:46: Exactly.

00:12:47: Which brings us a really nice closing thought From Ross Griffin and Ricardo Kahn

00:12:51: That innovation isn't a department.

00:12:53: It's a mindset.

00:12:54: you can't just hire ahead of innovation and expect magic, the real competitive advantage is still competent engaged people who feel safe enough to try a better way.

00:13:03: it really does all come back to The People doesn't?

00:13:05: Whether its vibe coding empowering an engineer or tact planning creating a sustainable rhythm for the trades All this tech is supposed remove friction so humans do their jobs better.

00:13:17: That's the synthesis of these last two weeks, I think.

00:13:19: The tools, AI and twins are getting there but industry is waking up to a fact that our processes, contracts or management styles have to evolve too!

00:13:28: You can't run twenty-twenty six tech on an nineteen ninety contract.

00:13:32: Well said you need design the operating system before installing the app.

00:13:36: Perfect analogy.

00:13:37: If you enjoy this episode new episodes drop every two weeks.

00:13:41: Also check out other additions in Smart Manufacturing and Digital Power Tools.

00:13:45: Thanks for lending us your ears and we'll catch you on the next deep dive.

00:13:48: Keep learning!