June 18, 2025 • 45 mins
Embark on a transformative journey with Hugh Seaton, CEO of The Link, as we traverse the changing landscape of construction technology. Having worked across continents and industries, Hugh brings a wealth of experience to the table, from his early days in Hong Kong and Taiwan to reshaping the construction world. This episode peels back the layers of traditional construction practices to reveal how data-driven specification management and forward-thinking IT strategies are revolutionizing the industry. Armed with a case study from The Link, Hugh illuminates how actionable data can streamline construction processes, moving us beyond the paper-laden methodologies of the 1970s.
As we venture further, the conversation turns to the sophistication of AI and its burgeoning role in construction. Language models like GPT-3 are under the spotlight, with the RICE Framework guiding us through crafting precise inquiries to elicit the most relevant information. Within the industry's technological tapestry, we find innovation teams working diligently to integrate new tools and BIM, changing the face of project delivery and embracing the strategic side of IT beyond mere support. Hugh shares how these expert teams are not just adopting but also selectively tailoring software, ensuring that the pace of technological change matches the industry's unique demands.
Looking toward the horizon, Hugh and I contemplate a future where learning and technology are inextricably linked within construction. From AI's potential in generative design to the liberation of the workforce from mundane tasks, we discuss an industry on the cusp of a new era. Technology's role in providing real-time, on-the-job training promises to revolutionize how we approach building and design. Join us for an episode that not only reveals the current state of construction technology but also charts a course for its exciting future.
Contact the Future Construct Podcast Produced by BIM Designs, Inc!
- BIM Designs, Inc.: minority-owned, US-based, union-signatory preconstruction technology firm, offering turnkey BIM modeling, laser scanning, coordination management, and other VDC solutions to the AEC industry.
- Schedule a free consultation: sales@bimdesigns.net.
- Subscribe to our weekly blog and our Future Construct Podcast
- Suggest a podcast guest
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:11):
Welcome builders and innovators to another episode of
the Future Construct podcast.
I'm your guest host, mark Oden,and today we have the privilege
of diving into the world ofconstruction technology with a
true visionary in the field.
Our guest today, for the thirdtime on our show, is none other
than Hugh Seaton, the CEO of theLink, a software company that
combines cutting-edge datascience and artificial
(00:31):
intelligence to makeconstruction documents
consumable for contractors.
They've spent years testing andrefining to ensure that teams
get the information they needwhen they need it, in the most
useful form possible.
Hugh is a seasoned professionalwho has dedicated his career to
uncovering the core truths indiverse realms, from navigating
the intricacies of advertisingmetrics to understanding
(00:52):
business culture in China andtackling the complexities of
startup financing.
Hugh first made an appearance onthe Future Construct podcast on
season one, episode 14, wherethe conversation centered around
the future of AR and the techevolution.
His second appearance was onseason three, episode three,
where Hugh discussed enablingsoftware and internal systems to
easily translate data.
(01:12):
Hugh's current passion lies ina domain that's relevant to all
of our listeners theconstruction industry.
As the general manager of theCrosswalk business line at CSI
and Symmetry, hugh gainedinvaluable insights into the
fundamental workings ofconstruction.
He delved deep into howcompanies leverage data to drive
their projects forward.
(01:33):
Not stopping there, hugh wearsmultiple hats in the
construction realm.
As the host of the ConstructedFutures podcast, he engages in
thought-provoking conversationswith two construction rock stars
every week, offering a glimpseinto the future of the industry.
His expertise doesn't end withdiscussions.
Hugh authored the ConstructionTechnology Handbook, providing a
much-needed primer ontechnology for today's
(01:55):
construction professionals.
So join me in this fascinatingconversation as we explore
Hugh's journey, the future ofconstruction technology and the
groundbreaking solutions thatthe link brings to the table.
Welcome back to our show andthank you for all of your many
contributions in the AEC space,hugh.
Thanks, mark, it's great to behere again.
So happy to have you To getgoing.
I'd love to dive into yourbackground even more than the
(02:17):
past couple of episodes.
In your first feature on ourshow, you spoke about growing up
in the 80s and spending a lotof time in Hong Kong and Taiwan
before moving back to the US andworking in advertising within
the technology space.
What experience or momentcatalyzed you to pursue a career
and technology to begin withyou?
Speaker 2 (02:34):
Yeah, when I first came back from round one of
China in 1998, I got a job withPhilips Electronics the ad
agency for Philips Electronicsand I've always been a bit of a
nerd.
When I was a kid I studiedlasers and studied electronics
and, like a lot of people, triedto make some things with
(02:54):
electronics and succeeded.
So it sometimes failed withothers.
But it really lit up when I wasable to work with Philips
Electronics I just saw a lot ofwhat they were doing.
They were actually partneredwith the old Bell Labs, a
company called Lucent, so wewere seeing some pretty cutting
edge stuff and that just got mereally excited and it's
continued ever since.
Speaker 1 (03:12):
That's amazing.
You've had a diverse career,delving into areas such as
advertising, metrics and, asmentioned, the business culture
of China start-up financing,before focusing on construction
industry.
How have these differentexperiences shaped your approach
to analyzing and sharinginformation about the
construction sector here?
Speaker 2 (03:29):
Yeah, I mean what I'm doing now isn't so different
from what I did.
The very first job I had wastranslating specifications from
English into Mandarin inSouthern Taiwan.
So there's been a through lineof how do people make better
decisions, how do people createLike, how do you brief somebody,
how do you create theconditions for someone to do
(03:51):
their job really, really well,and that's a lot of what
specifications are.
You know, sometimes that tookthe form of briefing a creative
team.
Sometimes that took the form ofmanaging a team.
There's been a bunch of waysthat that's shown up, but I've
got to be honest, I lovewatching people do great things.
I love watching people createamazing stuff, whether it's a
(04:15):
business plan, it's a product,it's something creative in the
advertising world, it's justsomething about human capability
that I really am excited about.
And, in fact, my last startupprior to this one, we did a lot
with e-learning, and a lot of itwas that A lot of what
e-learning is supposed to bedoing is helping adults to
become better than they were atwhatever it is they're
(04:36):
interested in.
So there's a bit of a throughline in really loving to create
the conditions for people to dowhat they do best.
Speaker 1 (04:45):
Yeah, let's talk about your current startup, the
link.
So it's, from what I understand, a construction specifications
management platform that pullsdata and insights from
construction specifications forproject teams and, as you
mentioned, it seems to be verywell connected to your past.
Can you walk us through aspecific example or case study
where the link has significantlystreamlined the construction
process for a project team?
Speaker 2 (05:07):
Yeah, let me talk about the way it all started.
So I was lucky enough to berecruited to go work for CSI
Actually, the company that heldmy W-2 is supplier to CSI, a
company called Sympatry, anamazing group and I was asked to
go and help them to grow andcontinue to create their
(05:31):
technology and their softwareoffering.
That gave me a lot ofvisibility into a few things.
One of them is some of thestandards that they use, but the
other one was howspecifications have evolved over
time and how they kind ofhaven't evolved as other things
have.
So specifications inconstruction right now are still
(05:51):
documents.
Fundamentally they behave thesame way they did in the 70s
when they were being courieredaround to each other.
Yes, they're PDFs, yes, theyget emailed and so on and so
forth, but there isn't the samereason for them to be all in one
document quite like there was.
For a number of reasons it'sgoing to stay that way for a bit
, and one of them is just legalthey're all in one place that
(06:13):
someone can stamp and put theirreputation behind.
There are ways in software to doit differently than that, but
that's one thing at a time.
So, to answer your question,you wind up getting especially
in a classic design bid buildproject, but even in design,
build and a number of otherproject delivery methods, you
get a bunch of specs all at once.
(06:34):
The project team has to onboardthese specs really quickly
understand them, start to writecontracts for subcontractors and
so on, and it's really not thateasy.
It's talking about 2,500 to4,000, sometimes more pages, a
pretty dense, legalese text notalways legalese, but certainly
technical text.
That's hard for people tonavigate.
(06:55):
It's hard for them to getthrough it in a way that is
coherent.
They manage, but they usuallymanage by putting bodies on the
problem and by getting it,getting through it pretty
painfully.
So what we do is the very firstthing is just to do.
Some of the outcomes early inthe project that you want are
like a submittal log and someother things.
So the first thing that we dois make the onboarding of the
(07:19):
spec faster and easier.
So the next piece, though, mark,that more clearly answers your
question, is the idea that afterthe beginning of the project,
people tend to look at the speca lot less than they did In the
beginning.
You don't have a choice.
You have to go through andunderstand what you're required
to do and what you're on thehook for.
But later, once you're building, people have a tendency to look
(07:41):
at the drawings a lot more thanthey do the specs.
The problem is someone saidthis once and I love it is
lawyers don't read drawings,they read specs.
So at the end of the day, ifthere's a claim or a problem,
they're going to go to the specs, not the drawings.
So what we do now is we helppeople to go from this older
kind of mentality where it'salmost like an old map right,
(08:03):
like a Rand McNally map that youonly use if you're going really
far or if you're lost in themiddle of a desert.
And the analogy in constructionis if it's a big deal or it's a
disaster, okay, we'll gothrough the specs.
But otherwise people on theproject know how to build, they
know how to pour concrete, theyknow how to install windows.
So the temptation is to go withmuscle memory and then things
get caught expensively andredone with rework or they don't
(08:26):
get caught and then you have aclaim or even a lawsuit.
So what we look to do, and whatwe've been doing, is going from
that paradigm of an old map toGoogle Maps Now it's so easy or
ways or whatever you name thedigital map.
But the point is, if you're evena little unsure, you pull it up
and go check and that's the bigshift that you see.
(08:47):
In for project teams is asuperintendent, a PE, the PM.
If they're even a little unsure, they're like yeah, I'm pretty
sure I know this, but thebarrier is so low to just ask,
because our system will then gointo the specs, find the answer,
get it to you in.
You know two seconds right, andwhat that allows you to do is
really reduce the risk of peoplegoing on autopilot or going,
(09:09):
more accurately, on musclememory, and instead they can go
consult the contractual documentand say, all right, we're sure
we're right, we're covered.
Speaker 1 (09:19):
That's awesome.
I'm so curious about it.
Can you speak a little bitabout you know training the
model or using LLMs, and youknow the data set and all of
that, yeah.
Speaker 2 (09:30):
And we do use LLMs.
Specifically right now we'reusing OpenAI GPT-4 plus, but you
know, every 15 minutes thereseems to be another one.
But that's okay.
What we didn't do is train amodel.
What we do is take the model asit is and fine tune it, because
that space is moving so fastand there's no way we're going
(09:51):
to I mean, they're raisingbillions of dollars.
So the state of the art in AImodels is not something a
startup at our size is going toengage in.
What we're engaging in is how dowe take these emerging
capabilities and make themuseful, make them into products,
because chat GPT or, yeah, chatGPT is interesting and it's a
(10:11):
generic problem, like Google is.
But if you want a specificproduct, something that has
security, something that hasworkflows around it, you need to
build actual software.
So that's what we focus on.
An example of that is most PMsand superintendents often
reference the last job or twojobs ago when they're just
thinking about their current job.
What we allow them to do isupload those specs to and point
(10:33):
spec GPT at that also, so theycan find the thing that there's
on the tip of their tongue orwhat did we use in that last one
.
Well, they can go find it likethat and then continue to use
past experience to influence andreally help them to execute the
project at the right minutelevel.
Speaker 1 (10:51):
I love that you and I'm super excited about diving
even deeper with you.
Would you highlight some uniquefeatures that the links, that's
apart from other specificationmanagement platforms?
If any others exist, I'd loveto hear.
Speaker 2 (11:02):
Well, actually, let me start with the last half of
your question, because that's areally good place to start.
People have spent a lot of timemaking drawings better, like
whether it's planned grid orit's pro-core.
A bunch of things have beendone on the drawing and on the
modeling side, but specs havereally not had a lot of
attention.
There's a couple of pretty goodspec writing platforms, but
(11:25):
they're really more aboutwriting them and not so much
about managing.
They're really more for thearchitect side of the contracts,
so to speak, as differentproject delivery methods have
grown.
The fact is, you don't alwaysget all of the specs at once and
they may change and things youknow grow, design build being an
easy example of that.
You get about a third of yourdrawings and you get or of your
(11:48):
specs, excuse me and you getgoing and then you fill it in
over time.
So one of the things that wethink is absolutely missing is
the ability to manage specs asthey change, as they get amended
, but also just the fact thataccessing them and effectively
using them it's a really archaicsystem that people have right
(12:10):
now, whether it's a viewer,whether it's even if you get a
submittal log, that's not thewhole spec.
That's just the submittals.
I'm getting a little deep intomy own story here.
So the net net is we think thatit's really a ripe time for not
just a platform that can managespecs on the contractor side,
(12:33):
but also really do we need tohave them as documents?
These are collections of datathat relate to each other and
we've been doing databases, youknow, everywhere for 40 years
now.
Turning those documents intodata is sort of the fundamental
thing that we look to do at thelink, and we actually spent a
bunch of time with a guy whoworks for the Air Force I
(12:55):
believe excuse me, his PhD isbeing funded by the Air Force
who does taxonomies andontologies.
What that means is he helps usto look at specs and start to
create buckets so you can breakthem down into data.
Whenever the project is, it'llall be the same kind of buckets
(13:15):
Took a lot of work.
We put thousands of specsthrough it and that's the
beginning of us starting theprocess of changing what is
fundamentally a document intofundamentally, a database,
because that's what it should be.
If you think about what BIM is,bim was originally meant to be
it still is meant to be adatabase of information about a
building.
That is, of course, related toa geometric model, but the point
(13:38):
is it was meant to be data, notjust geometry, and similarly,
specs are supposed to be thenarrative of the project what
you're building, how you'reinstalling it, how you're
preparing before installation,how you're inspecting it, how
you're paying for it, you know,how you're meeting about it, all
this sort of thing.
There's no reason that needs tobe all in a document, and one
(13:59):
of the problems we have Imentioned before these are long
documents.
Part of the problem is you needto include everything that's
going to be needed for atwo-year project into the same
document, which means, at anygiven moment, you only need you
know 5% of what's in that doc,probably less than that.
Again, it's a very archaic wayof doing things.
It assumes that you have a setof documents that get stamped
(14:21):
and that's the legal agreement.
Well, you can do things.
You don't even need blockchain.
Just normal, you know, justnormal software.
You can see who did what.
You can audit what was changedwhen it was changed, way more
efficiently than a PDF that hasyou know markup on it, because
you can't do a lot with a markup.
You certainly can't do it atscale.
Someone has to physically lookat it.
(14:43):
So the point I'm making is youknow you started this with a
great question about specmanagement platforms and the
first thing to do to make thesethings more manageable is to
kind of change the way youinteract with them, and that's
the first thing we did.
So we pointed in LLM at them,and now you can.
The way you interact is nowmuch more conversational, much
(15:04):
more directed, much moreintelligent.
But then, downstream, what youneed to do is say great, these
documents, for reasons that arebeyond this podcast, they're
probably going to stay beingproduced the way they're
produced for the next littlewhile.
But once they're handed over tothe project team, why do they
have to stay a document?
Why can't we turn them intodata and then do all the
wonderful things we know how todo with data, whether you're
(15:25):
searching, recombining things,networking them with other
information like standards andRFIs, and change orders and
product information and so on.
We have a whole network ofinformation that goes into the
building of a product or abuilding of a building, and
specs often reference otherthings.
But if we turn them into data,you don't need to just do the
(15:47):
reference anymore.
You can have it all in oneplace, and that's a long-winded
way of saying that.
We're on the beginning of ajourney where we think there's a
new way to think about specsand the sort of building
requirements that they're a partof, so that it's a little bit
less siloed.
You're asking a lot less ofpeople when they're busy to find
(16:08):
things and execute against them, but also, over time, what
happens when regulations change?
What happens when you're beingasked what's the carbon content
of this building?
Right now it's hard to dobecause you have to go hunt
through data.
Sorry, you have to go huntthrough documents.
Well, that should be a quicksearch that should be very fast
to say here we have this EPD.
(16:28):
These are documents that comefrom the environmental EPA as
well as building productmanufacturers.
I ought to be able to referenceEPDs automatically just based
on the fact that I have data andI have this EPD coming in.
It shouldn't need to bedocument against document.
That's just one example of howwe see this changing.
(16:49):
Again, once you have things inthe form of data, it unlocks a
ton of downstream innovation andability to do smart things with
your project.
Speaker 1 (17:02):
Yeah, really, I can just see so many use cases and
simplifying the interaction withthe data or the specs and where
it makes so much sense tocreate a spec GPT using an LLM
like chat GPT and something theaudience may be aware of chat
(17:22):
GPT has been.
There's a common word calledhallucinations, where it may
just create an output on aninput.
So I'm wondering how are youtackling hallucinations when it
comes to spec GPT?
Speaker 2 (17:35):
Yeah, that's a great question.
There's a number of ways thatyou answer that.
The very first one is theunderlying models themselves are
getting better all the time.
So the OpenAI's and the Metasand the Mistros in France and
Google, of course, they wereaware of this time last year, so
there's been a lot of work tomake the models better and
(17:56):
better.
In fact, what isn't commonlytalked about is prior to
November 2022, when chat GPTreally broke out into the scene.
The underlying models have beenaround for a couple of years.
Gpt-3 had been around for atleast a year.
The problem was it produced arange of answers that weren't
that useful.
The big innovation that madeall this happen is OpenAI, among
(18:18):
others, innovated a way oftraining them to only give you
the best answer.
So an extension is calledreinforcement learning with
human input.
The extension of that is tocontinue to train the models to
give answers that they know camefrom somewhere, as opposed to
they're just filling in whatlooks like the best answer.
(18:40):
So the first line of defense isthe underlying models continue
to get better.
In fact, gpt-4 plus or turbo Ibelieve it's turbo.
It was out last week, so, andif this podcast comes out in a
month.
That's probably still going tobe true because it came out with
another one.
They're just doing it soquickly.
That's point one.
Point two is asking betterquestions.
(19:01):
So if you treat spec, gpt, chat, gpt, any one of the LLMs as if
you're talking to Google,you're going to get much more
variable answers than if you aska better structured question.
Typically, you want to beseparating out parts of the
question like who are you askingas?
What's the context?
(19:21):
What are you looking for?
Are there any examples?
Depending on what you're askingfor, we bake that.
You don't see that as the user.
Well, we bake that into the wayyour question gets asked.
So you type in your questionbut we fill in some of the other
things so it'll answer theproblem in a more narrow way.
A general rule of thumb for allAI, whether it's deep learning
(19:42):
or whether it's LLMs or even thevision models the smaller you
make a problem, the narrower youdefine the problem, the better
it's going to be able to answerthe question and, by the way,
that's true for humans too.
One of the most famous adminever said give me the freedom of
a tight brief, and his pointwas the better you define this
is David Ogily.
The better you define a problem, the more I'm able to put
(20:03):
resources against solving thatinstead of trying to figure out
what you meant.
So the same thing is true withLLMs If you can make the problem
smaller by defining it, bysaying who's the persona, what's
the thing you're looking for?
Is there any context that helps?
Are there any constraints Ishould know about, and do you
have an example?
That's actually a frameworkcalled the Rice Framework
R-I-C-C-E.
(20:25):
That is just a better way ofwriting prompts.
In two weeks I'm actually doingthe whole webinar just on how
to write prompts, because Ithink the industry and just
everyone in the economy is on apath of learning how to interact
with LLMs more effectively, notas if you're talking to
software and not as if you'retalking to Google, but an
(20:48):
entirely new class of cognitivesystems that are a little way
more powerful than what we'reused to, but they can also be a
lot more variable than whatwe're used to.
So, just like if you weretalking to a human, the better
you brief them, the better youranswer is going to be.
So hallucinations are just likea human.
They will still make mistakes,but the degree to which those
(21:09):
mistakes are made is becomingmuch less than it was.
But the final way we deal withthat is by giving people access
to where the answer came from.
So in our interface you canclick on a button and see the
part of the spec that thatanswer came from.
So that's the ultimate point isgo look for yourself.
Speaker 1 (21:28):
That's brilliant.
Thank you, Hugh.
I appreciate that and this iscertainly a shared passion
between you and I, and I'm sureyou and I can and we'll talk for
hours on this.
To transition out of the linkspecifically and out of spec GPT
, We'd just love to talk alittle bit more broader sense.
What do you feel are thecurrent technological challenges
in the construction space,above and beyond the link that
(21:52):
you've been looking at?
Speaker 2 (21:53):
Yeah, I think there's some things everybody knows
about, and that is BIM should bemore a part of how projects are
managed.
I think ProCore has done apretty good job of bringing that
into the field.
They're not the only ones tohave done it, they're just
seeing a lot more of that.
I think construction has had apretty heavy couple of years
(22:16):
where a lot of money and a lotof options have been presented
to contractors and along the waythey got pretty good at
assessing and integrating andinternally marketing new ideas.
So you've seen, even since Iwrote the book, the growth of
(22:36):
innovation teams.
They're not always called that,but let's just for the moment
say that you know a team whosejob it is to survey what's out
there but also survey what'sneeded by their.
So constantly talking toproject teams to find out what
they're missing, what they coulddo with, what processes they
could do better and trying tomatch those and then, once
they've done a couple of pilotsand worked out kinks, then
(22:56):
figuring out how to market itinternally so that it gets
adopted, because it's prettyinfrequent that some central
authority tells everybody youabsolutely have to use this.
That happens for things thatyou don't have a choice, like
the accounting software or yourproject management software, but
usually that those are most ofthe time those are the only two.
I mean, sometimes it depends onthe contractor.
(23:16):
So I think that you know a lotof things are going better than
they were.
I will tell you.
I think the biggest opportunityfor the industry totally is for
IT teams to get elevated to astrategic role, so to have a
real CIO who is C-suite, notjust the IT director with a
better title, and that there's.
(23:37):
You're seeing that.
You know Albert Ricci did agreat job pretty early.
Actually, they're not at allalone.
Dpr is famous for being good atthis.
You know Suffolk has beenfantastic at this.
I'm going to forget to mentionsome folks, our friends at Ellis
Don, who I've got a lot ofaffinity for.
They were wise enough to chooseus as part of their accelerator
.
The degree to which they have afully functional internal
(24:01):
software team is surprising.
It's pretty amazing, and youknow they spent a lot of time
and a lot of money getting there.
But the point I'm making is Ithink across the industry,
contractors need to spend moremoney on IT, because that's how
they're going to unlock a lot ofthe value that things like LLMs
and the link or you name it,because if we are alone helping
(24:23):
a process, that's great, but ifwe're part of an integrated set
of processes, the value we canadd is literally exponentially
more than it was.
So I think that's going to bethe big one.
In all honesty, cyber securityis such a thing right now that
I'll bet you a lot of them arebreathlessly trying to keep up
with the threats they face.
So, as much as they probablywant to be more strategic than
(24:43):
they are, I understand why it'sslower than everybody wishes it
were.
You know, the problem with thecontractor is unlike some other
fields of endeavor or otherprofessions if your documents
get locked up, it's reallyreally hard to proceed.
So the more digital they are,ironically, the more that's true
.
So they're getting really goodat fighting against ransomware
(25:07):
and some other attacks, butnevertheless I have a feeling
that's slowing things down.
But I think over the comingyears you're going to see the
role of IT and CIOs become morestrategic than has been true in
the past and probably beseparated so that there's an
infrastructure team who makessure that the light stay on, the
emails work and the ransomwaregets taken care of, and then the
(25:28):
second half.
That is really looking forwardto how these technologies can
help manage risk and deliverprojects.
Speaker 1 (25:37):
Thanks, and you did mention some industry titans
there that are settingincredible examples to all of us
, so testament to theirinvestments and their forward
thinking.
So, looking at your formalposition as a general manager of
the crosswalk business line atCSI and Symmetry, and having
gained the firsthand experienceat a fundamental level of the
(25:58):
construction industry, can youshare specific instances where
the industry has given you aunique insight into how
companies utilize data to runprojects?
Speaker 2 (26:11):
Yes, I think it's a very, very big question.
I think that there's a fewthings.
One insight that maybe iscounterintuitive to the question
you asked is what constructionteaches you is the limits of
what data can do versus humanintuition and judgment.
So you really are findingyourself saying the job of data
(26:36):
isn't to make decisions, itisn't always even to define what
the decision is.
It's to support the decisionfor sure and help guide the
intuition of someone who's seenthis problem 40 times before,
and often it's to make sure thatthe person realizes this isn't
the 41st time.
It's actually either new orit's different.
So I think that what you see inconstruction is because of the
(26:58):
complexity of what's reallygoing on and all the
interconnectedness of it.
So if there's a thing that Iwould say was really a
revelation for me, is theincredible interconnectedness of
construction, and ask anyonewho's tried to manage a
construction schedule.
They'll tell you that it's so.
I mean, every day there'sanother ding on it and it moves
here, and it moves there, and,oh my gosh, we have to send it
(27:20):
out to the subs, and so on.
That speaks to the fact thatone thing is late because
somebody has a flat tire and allof a sudden it has 15 impacts
that you have to then manage.
That sets an upper limit onwhat certain types of data can
help you with, because thatcomplexity means a human has to
kind of manage it and maketrade-offs, because it isn't
just that you move one thing andthen everything moves
(27:42):
accordingly.
You're now making trade-offs.
Am I going to accept a littlemore risk by having two trades
in the same room with, or threeor whatever?
Am I going to accept risk oftrade stacking?
Or am I going to accept risk ofdelivery time by not doing that
?
There isn't really an equationfor that.
Data doesn't get you there.
What data does do is start tohelp with benchmarking, help
(28:04):
people to understand how high isup and what is good and what
actually we should be doingbetter.
Then I think it helps people tounderstand the problem a little
bit better, and definitely lesson the decision side than on
the operation side.
The more visibility people haveto what's going on, the better
they can apply that intuitionand say we're going to do better
(28:24):
than this.
An example of that is a numberof companies have gone and
started to automate.
Things like this is a littlemore on the trade side, but
things like the fab shop.
There's a lot of the same thinggoing on.
It's not quite a manufacturingwell, it is manufacturing
technically, but it's not likean assembly line.
Nevertheless, there's enoughrepeatability that you can start
to really tighten up how you'redoing what and how you're
(28:46):
making what decisions.
I think the number one placethat you see data helping
construction companies isvisibility.
The next thing after that isstarting to really think about
process and process redesign.
Frankly, I don't see a ton ofthat Seeing it more in the fab
shops, a little less elsewhere,but it's happening.
The final one that is the holygrail is predictive analytics.
(29:09):
I think that predictiveanalytics in construction is not
impossible.
It just takes more creativityto think about what you're
predicting and what's it reallytelling you.
I think what it's not going todo is tell you how things are
going to work out.
It tells you where the risksmight be and what you should pay
attention to.
You're seeing some folks thatare making some headway there.
(29:30):
The problem is, as always,finding the same data that you
can aggregate together into bigenough numbers to run math
against it, much less train.
A deep learning model is alwaysgoing to be tricky.
You really have to ask yourselfwhat are we training?
How complicated is the thingwe're trying to predict?
I mentioned before a generalpoint with AI is the more you
(29:54):
can narrow the problem, thebetter it's going to perform.
Frankly, if you think aboutthings like surveys, if I want a
survey to be valid, I want tohave a big group of people and I
want to have as few questions,as few cross tabs as possible.
It's a little like that themore of the same thing you have
against the narrower problem,the more accurate it's going to
be.
Well, that's a challenge fordata for us.
(30:16):
Just to recap again, I thinkthe first thing is visibility.
That makes the existingintuition and experience that
much more valuable.
Then, after that, you get theprocess redesigned and you
ultimately get to somepredictive analytics.
It's all coming one way or theother.
Speaker 1 (30:33):
Yeah, thanks so much.
You're definitely highlighting.
One of the largest challengesin terms of training the LLM, or
predictive analytics, is how tofind enough data and how to
make sure that you're trainingthe model.
How do you handle finding thedata and ownership of data and
transporting the data?
Speaker 2 (30:50):
It's a bunch of things in there.
I'll tease apart.
The first one is the value ofan LLM.
Why are they so transformativeis you don't need to train the
model, you need to fine tune it.
So it goes and finds just thething you want.
It comes to you with literallytrillions of parameters already.
That's really transformative,because now, all of a sudden,
it's going to be useful forconstruction.
(31:12):
The way you fine tune it is youhave people go through the
exercises you want and point outto it it should have been this,
it should have been this, itshould have been this.
Now you're talking abouthundreds and thousands of data
points, which, while annoying,is absolutely doable.
That's point one is you want tomake sure you're selecting the
right model.
The truth is that the top endof them are all pretty good.
(31:33):
The next is do I need to finetune it?
If so, how do I create arepeatable process for people to
train it?
There's a couple of ways aroundthat.
You also talked about storingdata and how do you handle it.
It's a really big point, and Ithink you were also asking about
(31:58):
who owns it.
That's actually a bigger point.
Contracts right now are notwritten for this.
It's really something thatpeople are starting to get
smarter about, but the classicAIA contract didn't need to deal
with this Not really, becausethey didn't think about
derivative data.
The way the magic reallymatters right now isn't the
project data, the plans and thespecs and all that.
(32:20):
That's pretty clear cut.
It's funny that it often is thearchitect that owns it, but
that's not what you're askingabout.
While operating on that, I'mcreating new data and metadata
and so on.
That's still being figured out.
Frankly, I'm not sure thereshould be a standard there.
I think that should besomething that gets negotiated
depending on what both partieswant.
(32:40):
I think what has happened, whatI've seen, is contractors gave
up a lot of data four or fiveyears ago to start-ups who then
went on to raise huge rounds.
They felt a little bit taken in.
Wait a minute.
We gave you all this data,which wasn't free for us to
organize and produce.
We put effort and resourcesinto this.
We got to thank you from youand off you went into the sunset
(33:04):
.
Some of these companies didn'tsucceed, so it is what it is.
It did leave a bad taste incontractors' mouths.
Now they want to know eitherthey own it or there's some
compensation for the effort theyput in.
That's more than just payingfor hours.
How do we take part of thisupside?
I think the ownership thing isgoing to take a little while,
(33:25):
not least because, like I said,I think that should be
negotiated in the contract,depending on what the different
goals of the parties are.
But the other one is on thecontractor side.
You have very accomplished butvery focused lawyers who are
really good at constructioncontracts and have had no reason
to be good at IP law.
(33:45):
So a lot of them are strugglingright now to figure out how to
write an IP policy and how tothink that through.
And you know they're smartpeople that really understand
their business.
Probably don't love the idea ofbringing an IP specialist in,
but the funny thing about IP is,almost anywhere you go, whether
you're a startup or you're, youknow, a non-construction
corporation, IP is sospecialized that you very often
(34:08):
bring in an IP specialist, evenif you're, you know, general
counsel at Proctor Gamble,Because it's just a thing that
people really get good at.
So I think the IP question isjust starting to get asked and
properly thought through and Idon't think the I don't even
think we understand where theproblem is going to be in coming
years, because what if you'veautomated a process and now the
(34:31):
automated process is creatingmetadata?
Who owns that Like?
So I think that it's going toget a little squirrely, but
we'll figure it out.
And that's where, again, Idon't think a standard is a
great idea.
I just think it's going todepend.
Some contractors are going tocare, some owners are not going
to care.
Some owners are going to reallycare.
You know, if you're putting upone building a year, you care
less about the IP than if you're.
(34:53):
You know target.
You put up, you know, 50projects a year.
You see what I mean.
Like, for one group it might besomething that really matters
for their competitiveness andfor another it may not.
So I think it'll be an area fordiscussion.
Speaker 1 (35:10):
I love it, hugh, and you're always the best one to
start the discussion.
And, speaking of, you're thehost of the Constructed Futures
podcast.
We're on a weekly basis.
You host two rock stars in theconstruction industry and you're
the author of the ConstructionTechnology Handbook, which
provided a primer on technologyfor Didey's construction
professionals.
We'd love you to tell me alittle bit more about the
Construction Technology Handbook.
(35:31):
What was your inspiration andwhat did you find that helped
you transform constructionpractices along the way?
Speaker 2 (35:38):
Yeah, I think what I found was I wanted to write a
book because I wanted to divedeeper and dive deeper into the
industry and talk to the mostamazing people I could find, and
it was kind of early.
I was still pretty new to theindustry.
I'd done some things in thebackground, but not in the
middle of it, and the cool thingabout writing a book is almost
no one says no to a request fora book interview, so I got to
(36:00):
talk to some pretty amazingfolks.
I'm actually doing a secondbook, which, not surprisingly,
is called AI and Construction,so I'm interviewing a little bit
less this time.
I'm starting from a differentbasis.
The point of that book, though,and sort of the core insight of
the first one of theConstruction Technology Handbook
, was people talk over eachother too much.
(36:21):
So folks who spent 20 yearsfiguring out how to pull a
building out of the groundprobably don't spend a lot of
time wondering what API standsfor, and yet people are throwing
this word at them all the time.
So I actually spend the firstthird of the book defining terms
.
I mean, I literally define whatWi-Fi is Understanding.
Most people know what it means,but have probably never had a
(36:41):
discussion about it so thatpeople walk away saying, look, I
may have known some of that,but I know all of it now, or at
least I've heard all of it now,because I wanted to level the
playing field so that consumersof technology would be able to
make better decisions than ifthey have to kind of shut the
brain off and get through thepart with all the words they
don't know and then have to feela little stupid asking
questions because it's on themto know whether the thing
(37:05):
they're buying is what they need.
But if they're not armed withthe right terminology and
understanding of what's going on, they're really hands-strong
and that slows everything down,because now people are like,
look, maybe this would help me,but I don't understand it well
enough and I can't take a riskwith this project.
So if you give people at leastthe table stakes of what these
words mean and thinking abouthow software is built and so on,
(37:27):
they're better able to make adecision which helps both sides
it helps the contractor and ithelps the vendor of software and
it helps the people managingthe whole thing.
That was the big deal.
Speaker 1 (37:40):
So, in addition to I mean I'm really excited to hear
about this new book you have inaddition to the success of the
last one and the new book thatyou're working on, and I'm sure
it all ties together to thisquestion, which is what is your
vision for the future ofconstruction and how do you see
the work you're doingcontributing to the evolution of
the construction, bestconstruction practices and
project management?
Speaker 2 (38:01):
Yeah, I think that it's important not to get
focused on one part of what'sgoing on.
I think there's a lot of changeand there's a lot of new ideas
and a lot of new ways of doingthings that are being worked out
and tried out.
It is such an intricate as Isaid before such an
interconnected endeavor to builda building that it's really
(38:22):
hard to change one thing withoutit impacting others.
So they're all inching forward.
But I think what you're goingto see and what I, to the degree
I have a vision about this iswe start to differentiate
buildings and think about how tobuild them in the way that's
most appropriate to them.
So things like industrializedconstruction, which absolutely
relates to specifications andabsolutely relates to AI,
(38:44):
especially on the generativedesign side, but some buildings
are going to be better for thatand some are not, and apartment
building is probably better forthat, multifamily maybe less so,
or a factory definitely less so.
So you see what I mean.
Like, I think that we have atendency sometimes, when we're
talking about construction, totry to continue to lumping it
(39:05):
together, and I think you'regoing to find specializations
more and more be the case.
I think that the degree towhich a human needs to be doing.
Lots of the minutiae inconstruction, I think, is going
to go away.
If there were a vision, itwould be.
We need fewer people who aredoing more value at, more of the
(39:28):
intuition and thinking thatgoes into making a great
building, and less of the movingstuff around and less of the
paperwork and less of the stuffthat people typically hate now
anyway.
And the interesting thing,though, is often when you say
things like that, veterans inthe industry will tell you yeah,
but how are people going tolearn if they don't go through
and do 15 submittal logs?
Well, the answer is the same.
(39:51):
Technology that automates.
Some of this is fantastic atcreating little micro-trainings,
so whereas before there reallywasn't a ton of training unless
maybe you're talking aboutsafety, and even that is kind of
recent you're now able to havethe LLM or whatever the software
platform is, create littlesnippets of learning along the
way, so you're learning whatthey call learning in the flow
(40:11):
of work.
So the other thing I thinkyou're going to find is the
tools that automate and thetools that make it easier for us
to not have to do all theMimousha that, again, people
don't usually love will be thesame ones that are the source of
learning and the source ofupskilling people, not only to
where they would have been 20years ago, but, as things
(40:32):
continue to change, they're ableto be there to do what in the
learning world, we callscaffolding learning, so
supporting learning.
So you're not taking on too muchat once, you're just learning a
little piece each time, butpretty quickly you get really
good at whatever the new thingis.
So I think, rather thanautomation, necessarily meaning
that our skills go away, I thinkthere's actually opportunity to
(40:54):
be thinking and saying thesethings can also write learning.
So let's do that, let's havethem do that and deliver
learning in a way that'soptimized for whoever the
learner is.
I think you're going to see arevolution across manufacturing,
across construction, acrosspretty much every service
business, where learning in theflow of work it comes from the
(41:14):
tools you're using, notnecessarily some learning
management system that no onepays any attention to.
Speaker 1 (41:19):
Yeah, and these lessons that have been learned
are very costly to the businesstoo, right?
So if there's a way to avoidthat the cost in the learning I
think that would berevolutionary as well.
Speaker 2 (41:32):
One interesting point really quickly.
Almost every project managerI've ever talked to has a list
of lessons learned.
They often need a little bit oftranslation because they're
usually reminding themselves orthey're delivered by them to
someone else, but the point isthere are huge repositories
across the industry where peoplehave taken notes.
They've taken notes on projects, they've taken notes on specs,
on drawings, so on.
There's a lot of context outthere that these new AI tools
(41:58):
have the promise to be able todistill into lessons that make
some sense, especially when youcombine them with the work that
the International Code Counciland CSI and ASTM and some others
.
They have amazing repositoriesof information about how to
build a building and I thinkthat those things start to
become unlocked.
We're going to be able to thinkabout buildings in a much more
(42:22):
sophisticated, comprehensive waythan any one human's brain
could handle.
Speaker 1 (42:28):
So, looking into the future, as you and I just were,
I would like to transition intothe final question of the show
and the tradition of our podcast, future Construct.
If you could project yourselfout 25 years and wanted to have
any device or technology thatwould benefit you personally,
what would it be and what wouldit do?
Speaker 2 (42:45):
25 years is an awfully long time.
I want a spaceship that'll takeme to the moon, where I can
hang out.
I'm kind of coming up with thatone on the cuff.
It doesn't really relate justto construction because, frankly
, in 25 years I think the way webuild buildings will be, some
(43:05):
of it will be unrecognizable andsome of it will look a lot like
right now.
But me personally, yeah, I likethe idea of personal transport
that can I don't know about themoon, but that can take you out
into space, and I love the ideaof experiencing space and I
think I'm young enough thatthat's a reasonable expectation.
I'm not so sure about Mars.
(43:27):
I moved to Austin and say Inever want to be culled again
and I don't know that Mars isgoing to satisfy that.
Speaker 1 (43:34):
Well.
Thank you so much for your time.
It was an incredibleconversation with you, hugh.
Your journey from the 80s inHong Kong to becoming a
construction tech visionary istruly captivating.
The link specificationmanagement platform, especially
with spec GPT, seems like a gamechanger, streamlining processes
for project teams.
Your insights into the industry, challenges, your unique
(43:55):
offerings, such as yourconstructive future podcast and
the construction technologyhandbook and your upcoming book
they all provide a richperspective of looking ahead.
Your vision for theconstruction industry and the
links role in shaping the futureis insanely inspiring.
Thank you, hugh, for being athought leader and to our
listeners, stay tuned for moreinsights on the future construct
(44:16):
podcast.
Hugh, it's been an absolutepleasure having you on the show.
Thank you so much for your time.
A thanks for having me.
Absolutely Thank you.
Welcome builders and innovators to another episode of
the Future Construct podcast.
I'm your guest host, mark Oden,and today we have the privilege
of diving into the world ofconstruction technology with a
true visionary in the field.
Our guest today, for the thirdtime on our show, is none other
than Hugh Seaton, the CEO of theLink, a software company that
combines cutting-edge datascience and artificial
(00:31):
intelligence to makeconstruction documents
consumable for contractors.
They've spent years testing andrefining to ensure that teams
get the information they needwhen they need it, in the most
useful form possible.
Hugh is a seasoned professionalwho has dedicated his career to
uncovering the core truths indiverse realms, from navigating
the intricacies of advertisingmetrics to understanding
(00:52):
business culture in China andtackling the complexities of
startup financing.
Hugh first made an appearance onthe Future Construct podcast on
season one, episode 14, wherethe conversation centered around
the future of AR and the techevolution.
His second appearance was onseason three, episode three,
where Hugh discussed enablingsoftware and internal systems to
easily translate data.
(01:12):
Hugh's current passion lies ina domain that's relevant to all
of our listeners theconstruction industry.
As the general manager of theCrosswalk business line at CSI
and Symmetry, hugh gainedinvaluable insights into the
fundamental workings ofconstruction.
He delved deep into howcompanies leverage data to drive
their projects forward.
(01:33):
Not stopping there, hugh wearsmultiple hats in the
construction realm.
As the host of the ConstructedFutures podcast, he engages in
thought-provoking conversationswith two construction rock stars
every week, offering a glimpseinto the future of the industry.
His expertise doesn't end withdiscussions.
Hugh authored the ConstructionTechnology Handbook, providing a
much-needed primer ontechnology for today's
(01:55):
construction professionals.
So join me in this fascinatingconversation as we explore
Hugh's journey, the future ofconstruction technology and the
groundbreaking solutions thatthe link brings to the table.
Welcome back to our show andthank you for all of your many
contributions in the AEC space,hugh.
Thanks, mark, it's great to behere again.
So happy to have you To getgoing.
I'd love to dive into yourbackground even more than the
(02:17):
past couple of episodes.
In your first feature on ourshow, you spoke about growing up
in the 80s and spending a lotof time in Hong Kong and Taiwan
before moving back to the US andworking in advertising within
the technology space.
What experience or momentcatalyzed you to pursue a career
and technology to begin withyou?
Speaker 2 (02:34):
Yeah, when I first came back from round one of
China in 1998, I got a job withPhilips Electronics the ad
agency for Philips Electronicsand I've always been a bit of a
nerd.
When I was a kid I studiedlasers and studied electronics
and, like a lot of people, triedto make some things with
(02:54):
electronics and succeeded.
So it sometimes failed withothers.
But it really lit up when I wasable to work with Philips
Electronics I just saw a lot ofwhat they were doing.
They were actually partneredwith the old Bell Labs, a
company called Lucent, so wewere seeing some pretty cutting
edge stuff and that just got mereally excited and it's
continued ever since.
Speaker 1 (03:12):
That's amazing.
You've had a diverse career,delving into areas such as
advertising, metrics and, asmentioned, the business culture
of China start-up financing,before focusing on construction
industry.
How have these differentexperiences shaped your approach
to analyzing and sharinginformation about the
construction sector here?
Speaker 2 (03:29):
Yeah, I mean what I'm doing now isn't so different
from what I did.
The very first job I had wastranslating specifications from
English into Mandarin inSouthern Taiwan.
So there's been a through lineof how do people make better
decisions, how do people createLike, how do you brief somebody,
how do you create theconditions for someone to do
(03:51):
their job really, really well,and that's a lot of what
specifications are.
You know, sometimes that tookthe form of briefing a creative
team.
Sometimes that took the form ofmanaging a team.
There's been a bunch of waysthat that's shown up, but I've
got to be honest, I lovewatching people do great things.
I love watching people createamazing stuff, whether it's a
(04:15):
business plan, it's a product,it's something creative in the
advertising world, it's justsomething about human capability
that I really am excited about.
And, in fact, my last startupprior to this one, we did a lot
with e-learning, and a lot of itwas that A lot of what
e-learning is supposed to bedoing is helping adults to
become better than they were atwhatever it is they're
(04:36):
interested in.
So there's a bit of a throughline in really loving to create
the conditions for people to dowhat they do best.
Speaker 1 (04:45):
Yeah, let's talk about your current startup, the
link.
So it's, from what I understand, a construction specifications
management platform that pullsdata and insights from
construction specifications forproject teams and, as you
mentioned, it seems to be verywell connected to your past.
Can you walk us through aspecific example or case study
where the link has significantlystreamlined the construction
process for a project team?
Speaker 2 (05:07):
Yeah, let me talk about the way it all started.
So I was lucky enough to berecruited to go work for CSI
Actually, the company that heldmy W-2 is supplier to CSI, a
company called Sympatry, anamazing group and I was asked to
go and help them to grow andcontinue to create their
(05:31):
technology and their softwareoffering.
That gave me a lot ofvisibility into a few things.
One of them is some of thestandards that they use, but the
other one was howspecifications have evolved over
time and how they kind ofhaven't evolved as other things
have.
So specifications inconstruction right now are still
(05:51):
documents.
Fundamentally they behave thesame way they did in the 70s
when they were being courieredaround to each other.
Yes, they're PDFs, yes, theyget emailed and so on and so
forth, but there isn't the samereason for them to be all in one
document quite like there was.
For a number of reasons it'sgoing to stay that way for a bit
, and one of them is just legalthey're all in one place that
(06:13):
someone can stamp and put theirreputation behind.
There are ways in software to doit differently than that, but
that's one thing at a time.
So, to answer your question,you wind up getting especially
in a classic design bid buildproject, but even in design,
build and a number of otherproject delivery methods, you
get a bunch of specs all at once.
(06:34):
The project team has to onboardthese specs really quickly
understand them, start to writecontracts for subcontractors and
so on, and it's really not thateasy.
It's talking about 2,500 to4,000, sometimes more pages, a
pretty dense, legalese text notalways legalese, but certainly
technical text.
That's hard for people tonavigate.
(06:55):
It's hard for them to getthrough it in a way that is
coherent.
They manage, but they usuallymanage by putting bodies on the
problem and by getting it,getting through it pretty
painfully.
So what we do is the very firstthing is just to do.
Some of the outcomes early inthe project that you want are
like a submittal log and someother things.
So the first thing that we dois make the onboarding of the
(07:19):
spec faster and easier.
So the next piece, though, mark,that more clearly answers your
question, is the idea that afterthe beginning of the project,
people tend to look at the speca lot less than they did In the
beginning.
You don't have a choice.
You have to go through andunderstand what you're required
to do and what you're on thehook for.
But later, once you're building, people have a tendency to look
(07:41):
at the drawings a lot more thanthey do the specs.
The problem is someone saidthis once and I love it is
lawyers don't read drawings,they read specs.
So at the end of the day, ifthere's a claim or a problem,
they're going to go to the specs, not the drawings.
So what we do now is we helppeople to go from this older
kind of mentality where it'salmost like an old map right,
(08:03):
like a Rand McNally map that youonly use if you're going really
far or if you're lost in themiddle of a desert.
And the analogy in constructionis if it's a big deal or it's a
disaster, okay, we'll gothrough the specs.
But otherwise people on theproject know how to build, they
know how to pour concrete, theyknow how to install windows.
So the temptation is to go withmuscle memory and then things
get caught expensively andredone with rework or they don't
(08:26):
get caught and then you have aclaim or even a lawsuit.
So what we look to do, and whatwe've been doing, is going from
that paradigm of an old map toGoogle Maps Now it's so easy or
ways or whatever you name thedigital map.
But the point is, if you're evena little unsure, you pull it up
and go check and that's the bigshift that you see.
(08:47):
In for project teams is asuperintendent, a PE, the PM.
If they're even a little unsure, they're like yeah, I'm pretty
sure I know this, but thebarrier is so low to just ask,
because our system will then gointo the specs, find the answer,
get it to you in.
You know two seconds right, andwhat that allows you to do is
really reduce the risk of peoplegoing on autopilot or going,
(09:09):
more accurately, on musclememory, and instead they can go
consult the contractual documentand say, all right, we're sure
we're right, we're covered.
Speaker 1 (09:19):
That's awesome.
I'm so curious about it.
Can you speak a little bitabout you know training the
model or using LLMs, and youknow the data set and all of
that, yeah.
Speaker 2 (09:30):
And we do use LLMs.
Specifically right now we'reusing OpenAI GPT-4 plus, but you
know, every 15 minutes thereseems to be another one.
But that's okay.
What we didn't do is train amodel.
What we do is take the model asit is and fine tune it, because
that space is moving so fastand there's no way we're going
(09:51):
to I mean, they're raisingbillions of dollars.
So the state of the art in AImodels is not something a
startup at our size is going toengage in.
What we're engaging in is how dowe take these emerging
capabilities and make themuseful, make them into products,
because chat GPT or, yeah, chatGPT is interesting and it's a
(10:11):
generic problem, like Google is.
But if you want a specificproduct, something that has
security, something that hasworkflows around it, you need to
build actual software.
So that's what we focus on.
An example of that is most PMsand superintendents often
reference the last job or twojobs ago when they're just
thinking about their current job.
What we allow them to do isupload those specs to and point
(10:33):
spec GPT at that also, so theycan find the thing that there's
on the tip of their tongue orwhat did we use in that last one
.
Well, they can go find it likethat and then continue to use
past experience to influence andreally help them to execute the
project at the right minutelevel.
Speaker 1 (10:51):
I love that you and I'm super excited about diving
even deeper with you.
Would you highlight some uniquefeatures that the links, that's
apart from other specificationmanagement platforms?
If any others exist, I'd loveto hear.
Speaker 2 (11:02):
Well, actually, let me start with the last half of
your question, because that's areally good place to start.
People have spent a lot of timemaking drawings better, like
whether it's planned grid orit's pro-core.
A bunch of things have beendone on the drawing and on the
modeling side, but specs havereally not had a lot of
attention.
There's a couple of pretty goodspec writing platforms, but
(11:25):
they're really more aboutwriting them and not so much
about managing.
They're really more for thearchitect side of the contracts,
so to speak, as differentproject delivery methods have
grown.
The fact is, you don't alwaysget all of the specs at once and
they may change and things youknow grow, design build being an
easy example of that.
You get about a third of yourdrawings and you get or of your
(11:48):
specs, excuse me and you getgoing and then you fill it in
over time.
So one of the things that wethink is absolutely missing is
the ability to manage specs asthey change, as they get amended
, but also just the fact thataccessing them and effectively
using them it's a really archaicsystem that people have right
(12:10):
now, whether it's a viewer,whether it's even if you get a
submittal log, that's not thewhole spec.
That's just the submittals.
I'm getting a little deep intomy own story here.
So the net net is we think thatit's really a ripe time for not
just a platform that can managespecs on the contractor side,
(12:33):
but also really do we need tohave them as documents?
These are collections of datathat relate to each other and
we've been doing databases, youknow, everywhere for 40 years
now.
Turning those documents intodata is sort of the fundamental
thing that we look to do at thelink, and we actually spent a
bunch of time with a guy whoworks for the Air Force I
(12:55):
believe excuse me, his PhD isbeing funded by the Air Force
who does taxonomies andontologies.
What that means is he helps usto look at specs and start to
create buckets so you can breakthem down into data.
Whenever the project is, it'llall be the same kind of buckets
(13:15):
Took a lot of work.
We put thousands of specsthrough it and that's the
beginning of us starting theprocess of changing what is
fundamentally a document intofundamentally, a database,
because that's what it should be.
If you think about what BIM is,bim was originally meant to be
it still is meant to be adatabase of information about a
building.
That is, of course, related toa geometric model, but the point
(13:38):
is it was meant to be data, notjust geometry, and similarly,
specs are supposed to be thenarrative of the project what
you're building, how you'reinstalling it, how you're
preparing before installation,how you're inspecting it, how
you're paying for it, you know,how you're meeting about it, all
this sort of thing.
There's no reason that needs tobe all in a document, and one
(13:59):
of the problems we have Imentioned before these are long
documents.
Part of the problem is you needto include everything that's
going to be needed for atwo-year project into the same
document, which means, at anygiven moment, you only need you
know 5% of what's in that doc,probably less than that.
Again, it's a very archaic wayof doing things.
It assumes that you have a setof documents that get stamped
(14:21):
and that's the legal agreement.
Well, you can do things.
You don't even need blockchain.
Just normal, you know, justnormal software.
You can see who did what.
You can audit what was changedwhen it was changed, way more
efficiently than a PDF that hasyou know markup on it, because
you can't do a lot with a markup.
You certainly can't do it atscale.
Someone has to physically lookat it.
(14:43):
So the point I'm making is youknow you started this with a
great question about specmanagement platforms and the
first thing to do to make thesethings more manageable is to
kind of change the way youinteract with them, and that's
the first thing we did.
So we pointed in LLM at them,and now you can.
The way you interact is nowmuch more conversational, much
(15:04):
more directed, much moreintelligent.
But then, downstream, what youneed to do is say great, these
documents, for reasons that arebeyond this podcast, they're
probably going to stay beingproduced the way they're
produced for the next littlewhile.
But once they're handed over tothe project team, why do they
have to stay a document?
Why can't we turn them intodata and then do all the
wonderful things we know how todo with data, whether you're
(15:25):
searching, recombining things,networking them with other
information like standards andRFIs, and change orders and
product information and so on.
We have a whole network ofinformation that goes into the
building of a product or abuilding of a building, and
specs often reference otherthings.
But if we turn them into data,you don't need to just do the
(15:47):
reference anymore.
You can have it all in oneplace, and that's a long-winded
way of saying that.
We're on the beginning of ajourney where we think there's a
new way to think about specsand the sort of building
requirements that they're a partof, so that it's a little bit
less siloed.
You're asking a lot less ofpeople when they're busy to find
(16:08):
things and execute against them, but also, over time, what
happens when regulations change?
What happens when you're beingasked what's the carbon content
of this building?
Right now it's hard to dobecause you have to go hunt
through data.
Sorry, you have to go huntthrough documents.
Well, that should be a quicksearch that should be very fast
to say here we have this EPD.
(16:28):
These are documents that comefrom the environmental EPA as
well as building productmanufacturers.
I ought to be able to referenceEPDs automatically just based
on the fact that I have data andI have this EPD coming in.
It shouldn't need to bedocument against document.
That's just one example of howwe see this changing.
(16:49):
Again, once you have things inthe form of data, it unlocks a
ton of downstream innovation andability to do smart things with
your project.
Speaker 1 (17:02):
Yeah, really, I can just see so many use cases and
simplifying the interaction withthe data or the specs and where
it makes so much sense tocreate a spec GPT using an LLM
like chat GPT and something theaudience may be aware of chat
(17:22):
GPT has been.
There's a common word calledhallucinations, where it may
just create an output on aninput.
So I'm wondering how are youtackling hallucinations when it
comes to spec GPT?
Speaker 2 (17:35):
Yeah, that's a great question.
There's a number of ways thatyou answer that.
The very first one is theunderlying models themselves are
getting better all the time.
So the OpenAI's and the Metasand the Mistros in France and
Google, of course, they wereaware of this time last year, so
there's been a lot of work tomake the models better and
(17:56):
better.
In fact, what isn't commonlytalked about is prior to
November 2022, when chat GPTreally broke out into the scene.
The underlying models have beenaround for a couple of years.
Gpt-3 had been around for atleast a year.
The problem was it produced arange of answers that weren't
that useful.
The big innovation that madeall this happen is OpenAI, among
(18:18):
others, innovated a way oftraining them to only give you
the best answer.
So an extension is calledreinforcement learning with
human input.
The extension of that is tocontinue to train the models to
give answers that they know camefrom somewhere, as opposed to
they're just filling in whatlooks like the best answer.
(18:40):
So the first line of defense isthe underlying models continue
to get better.
In fact, gpt-4 plus or turbo Ibelieve it's turbo.
It was out last week, so, andif this podcast comes out in a
month.
That's probably still going tobe true because it came out with
another one.
They're just doing it soquickly.
That's point one.
Point two is asking betterquestions.
(19:01):
So if you treat spec, gpt, chat, gpt, any one of the LLMs as if
you're talking to Google,you're going to get much more
variable answers than if you aska better structured question.
Typically, you want to beseparating out parts of the
question like who are you askingas?
What's the context?
(19:21):
What are you looking for?
Are there any examples?
Depending on what you're askingfor, we bake that.
You don't see that as the user.
Well, we bake that into the wayyour question gets asked.
So you type in your questionbut we fill in some of the other
things so it'll answer theproblem in a more narrow way.
A general rule of thumb for allAI, whether it's deep learning
(19:42):
or whether it's LLMs or even thevision models the smaller you
make a problem, the narrower youdefine the problem, the better
it's going to be able to answerthe question and, by the way,
that's true for humans too.
One of the most famous adminever said give me the freedom of
a tight brief, and his pointwas the better you define this
is David Ogily.
The better you define a problem, the more I'm able to put
(20:03):
resources against solving thatinstead of trying to figure out
what you meant.
So the same thing is true withLLMs If you can make the problem
smaller by defining it, bysaying who's the persona, what's
the thing you're looking for?
Is there any context that helps?
Are there any constraints Ishould know about, and do you
have an example?
That's actually a frameworkcalled the Rice Framework
R-I-C-C-E.
(20:25):
That is just a better way ofwriting prompts.
In two weeks I'm actually doingthe whole webinar just on how
to write prompts, because Ithink the industry and just
everyone in the economy is on apath of learning how to interact
with LLMs more effectively, notas if you're talking to
software and not as if you'retalking to Google, but an
(20:48):
entirely new class of cognitivesystems that are a little way
more powerful than what we'reused to, but they can also be a
lot more variable than whatwe're used to.
So, just like if you weretalking to a human, the better
you brief them, the better youranswer is going to be.
So hallucinations are just likea human.
They will still make mistakes,but the degree to which those
(21:09):
mistakes are made is becomingmuch less than it was.
But the final way we deal withthat is by giving people access
to where the answer came from.
So in our interface you canclick on a button and see the
part of the spec that thatanswer came from.
So that's the ultimate point isgo look for yourself.
Speaker 1 (21:28):
That's brilliant.
Thank you, Hugh.
I appreciate that and this iscertainly a shared passion
between you and I, and I'm sureyou and I can and we'll talk for
hours on this.
To transition out of the linkspecifically and out of spec GPT
, We'd just love to talk alittle bit more broader sense.
What do you feel are thecurrent technological challenges
in the construction space,above and beyond the link that
(21:52):
you've been looking at?
Speaker 2 (21:53):
Yeah, I think there's some things everybody knows
about, and that is BIM should bemore a part of how projects are
managed.
I think ProCore has done apretty good job of bringing that
into the field.
They're not the only ones tohave done it, they're just
seeing a lot more of that.
I think construction has had apretty heavy couple of years
(22:16):
where a lot of money and a lotof options have been presented
to contractors and along the waythey got pretty good at
assessing and integrating andinternally marketing new ideas.
So you've seen, even since Iwrote the book, the growth of
(22:36):
innovation teams.
They're not always called that,but let's just for the moment
say that you know a team whosejob it is to survey what's out
there but also survey what'sneeded by their.
So constantly talking toproject teams to find out what
they're missing, what they coulddo with, what processes they
could do better and trying tomatch those and then, once
they've done a couple of pilotsand worked out kinks, then
(22:56):
figuring out how to market itinternally so that it gets
adopted, because it's prettyinfrequent that some central
authority tells everybody youabsolutely have to use this.
That happens for things thatyou don't have a choice, like
the accounting software or yourproject management software, but
usually that those are most ofthe time those are the only two.
I mean, sometimes it depends onthe contractor.
(23:16):
So I think that you know a lotof things are going better than
they were.
I will tell you.
I think the biggest opportunityfor the industry totally is for
IT teams to get elevated to astrategic role, so to have a
real CIO who is C-suite, notjust the IT director with a
better title, and that there's.
(23:37):
You're seeing that.
You know Albert Ricci did agreat job pretty early.
Actually, they're not at allalone.
Dpr is famous for being good atthis.
You know Suffolk has beenfantastic at this.
I'm going to forget to mentionsome folks, our friends at Ellis
Don, who I've got a lot ofaffinity for.
They were wise enough to chooseus as part of their accelerator
.
The degree to which they have afully functional internal
(24:01):
software team is surprising.
It's pretty amazing, and youknow they spent a lot of time
and a lot of money getting there.
But the point I'm making is Ithink across the industry,
contractors need to spend moremoney on IT, because that's how
they're going to unlock a lot ofthe value that things like LLMs
and the link or you name it,because if we are alone helping
(24:23):
a process, that's great, but ifwe're part of an integrated set
of processes, the value we canadd is literally exponentially
more than it was.
So I think that's going to bethe big one.
In all honesty, cyber securityis such a thing right now that
I'll bet you a lot of them arebreathlessly trying to keep up
with the threats they face.
So, as much as they probablywant to be more strategic than
(24:43):
they are, I understand why it'sslower than everybody wishes it
were.
You know, the problem with thecontractor is unlike some other
fields of endeavor or otherprofessions if your documents
get locked up, it's reallyreally hard to proceed.
So the more digital they are,ironically, the more that's true
.
So they're getting really goodat fighting against ransomware
(25:07):
and some other attacks, butnevertheless I have a feeling
that's slowing things down.
But I think over the comingyears you're going to see the
role of IT and CIOs become morestrategic than has been true in
the past and probably beseparated so that there's an
infrastructure team who makessure that the light stay on, the
emails work and the ransomwaregets taken care of, and then the
(25:28):
second half.
That is really looking forwardto how these technologies can
help manage risk and deliverprojects.
Speaker 1 (25:37):
Thanks, and you did mention some industry titans
there that are settingincredible examples to all of us
, so testament to theirinvestments and their forward
thinking.
So, looking at your formalposition as a general manager of
the crosswalk business line atCSI and Symmetry, and having
gained the firsthand experienceat a fundamental level of the
(25:58):
construction industry, can youshare specific instances where
the industry has given you aunique insight into how
companies utilize data to runprojects?
Speaker 2 (26:11):
Yes, I think it's a very, very big question.
I think that there's a fewthings.
One insight that maybe iscounterintuitive to the question
you asked is what constructionteaches you is the limits of
what data can do versus humanintuition and judgment.
So you really are findingyourself saying the job of data
(26:36):
isn't to make decisions, itisn't always even to define what
the decision is.
It's to support the decisionfor sure and help guide the
intuition of someone who's seenthis problem 40 times before,
and often it's to make sure thatthe person realizes this isn't
the 41st time.
It's actually either new orit's different.
So I think that what you see inconstruction is because of the
(26:58):
complexity of what's reallygoing on and all the
interconnectedness of it.
So if there's a thing that Iwould say was really a
revelation for me, is theincredible interconnectedness of
construction, and ask anyonewho's tried to manage a
construction schedule.
They'll tell you that it's so.
I mean, every day there'sanother ding on it and it moves
here, and it moves there, and,oh my gosh, we have to send it
(27:20):
out to the subs, and so on.
That speaks to the fact thatone thing is late because
somebody has a flat tire and allof a sudden it has 15 impacts
that you have to then manage.
That sets an upper limit onwhat certain types of data can
help you with, because thatcomplexity means a human has to
kind of manage it and maketrade-offs, because it isn't
just that you move one thing andthen everything moves
(27:42):
accordingly.
You're now making trade-offs.
Am I going to accept a littlemore risk by having two trades
in the same room with, or threeor whatever?
Am I going to accept risk oftrade stacking?
Or am I going to accept risk ofdelivery time by not doing that
?
There isn't really an equationfor that.
Data doesn't get you there.
What data does do is start tohelp with benchmarking, help
(28:04):
people to understand how high isup and what is good and what
actually we should be doingbetter.
Then I think it helps people tounderstand the problem a little
bit better, and definitely lesson the decision side than on
the operation side.
The more visibility people haveto what's going on, the better
they can apply that intuitionand say we're going to do better
(28:24):
than this.
An example of that is a numberof companies have gone and
started to automate.
Things like this is a littlemore on the trade side, but
things like the fab shop.
There's a lot of the same thinggoing on.
It's not quite a manufacturingwell, it is manufacturing
technically, but it's not likean assembly line.
Nevertheless, there's enoughrepeatability that you can start
to really tighten up how you'redoing what and how you're
(28:46):
making what decisions.
I think the number one placethat you see data helping
construction companies isvisibility.
The next thing after that isstarting to really think about
process and process redesign.
Frankly, I don't see a ton ofthat Seeing it more in the fab
shops, a little less elsewhere,but it's happening.
The final one that is the holygrail is predictive analytics.
(29:09):
I think that predictiveanalytics in construction is not
impossible.
It just takes more creativityto think about what you're
predicting and what's it reallytelling you.
I think what it's not going todo is tell you how things are
going to work out.
It tells you where the risksmight be and what you should pay
attention to.
You're seeing some folks thatare making some headway there.
(29:30):
The problem is, as always,finding the same data that you
can aggregate together into bigenough numbers to run math
against it, much less train.
A deep learning model is alwaysgoing to be tricky.
You really have to ask yourselfwhat are we training?
How complicated is the thingwe're trying to predict?
I mentioned before a generalpoint with AI is the more you
(29:54):
can narrow the problem, thebetter it's going to perform.
Frankly, if you think aboutthings like surveys, if I want a
survey to be valid, I want tohave a big group of people and I
want to have as few questions,as few cross tabs as possible.
It's a little like that themore of the same thing you have
against the narrower problem,the more accurate it's going to
be.
Well, that's a challenge fordata for us.
(30:16):
Just to recap again, I thinkthe first thing is visibility.
That makes the existingintuition and experience that
much more valuable.
Then, after that, you get theprocess redesigned and you
ultimately get to somepredictive analytics.
It's all coming one way or theother.
Speaker 1 (30:33):
Yeah, thanks so much.
You're definitely highlighting.
One of the largest challengesin terms of training the LLM, or
predictive analytics, is how tofind enough data and how to
make sure that you're trainingthe model.
How do you handle finding thedata and ownership of data and
transporting the data?
Speaker 2 (30:50):
It's a bunch of things in there.
I'll tease apart.
The first one is the value ofan LLM.
Why are they so transformativeis you don't need to train the
model, you need to fine tune it.
So it goes and finds just thething you want.
It comes to you with literallytrillions of parameters already.
That's really transformative,because now, all of a sudden,
it's going to be useful forconstruction.
(31:12):
The way you fine tune it is youhave people go through the
exercises you want and point outto it it should have been this,
it should have been this, itshould have been this.
Now you're talking abouthundreds and thousands of data
points, which, while annoying,is absolutely doable.
That's point one is you want tomake sure you're selecting the
right model.
The truth is that the top endof them are all pretty good.
(31:33):
The next is do I need to finetune it?
If so, how do I create arepeatable process for people to
train it?
There's a couple of ways aroundthat.
You also talked about storingdata and how do you handle it.
It's a really big point, and Ithink you were also asking about
(31:58):
who owns it.
That's actually a bigger point.
Contracts right now are notwritten for this.
It's really something thatpeople are starting to get
smarter about, but the classicAIA contract didn't need to deal
with this Not really, becausethey didn't think about
derivative data.
The way the magic reallymatters right now isn't the
project data, the plans and thespecs and all that.
(32:20):
That's pretty clear cut.
It's funny that it often is thearchitect that owns it, but
that's not what you're askingabout.
While operating on that, I'mcreating new data and metadata
and so on.
That's still being figured out.
Frankly, I'm not sure thereshould be a standard there.
I think that should besomething that gets negotiated
depending on what both partieswant.
(32:40):
I think what has happened, whatI've seen, is contractors gave
up a lot of data four or fiveyears ago to start-ups who then
went on to raise huge rounds.
They felt a little bit taken in.
Wait a minute.
We gave you all this data,which wasn't free for us to
organize and produce.
We put effort and resourcesinto this.
We got to thank you from youand off you went into the sunset
(33:04):
.
Some of these companies didn'tsucceed, so it is what it is.
It did leave a bad taste incontractors' mouths.
Now they want to know eitherthey own it or there's some
compensation for the effort theyput in.
That's more than just payingfor hours.
How do we take part of thisupside?
I think the ownership thing isgoing to take a little while,
(33:25):
not least because, like I said,I think that should be
negotiated in the contract,depending on what the different
goals of the parties are.
But the other one is on thecontractor side.
You have very accomplished butvery focused lawyers who are
really good at constructioncontracts and have had no reason
to be good at IP law.
(33:45):
So a lot of them are strugglingright now to figure out how to
write an IP policy and how tothink that through.
And you know they're smartpeople that really understand
their business.
Probably don't love the idea ofbringing an IP specialist in,
but the funny thing about IP is,almost anywhere you go, whether
you're a startup or you're, youknow, a non-construction
corporation, IP is sospecialized that you very often
(34:08):
bring in an IP specialist, evenif you're, you know, general
counsel at Proctor Gamble,Because it's just a thing that
people really get good at.
So I think the IP question isjust starting to get asked and
properly thought through and Idon't think the I don't even
think we understand where theproblem is going to be in coming
years, because what if you'veautomated a process and now the
(34:31):
automated process is creatingmetadata?
Who owns that Like?
So I think that it's going toget a little squirrely, but
we'll figure it out.
And that's where, again, Idon't think a standard is a
great idea.
I just think it's going todepend.
Some contractors are going tocare, some owners are not going
to care.
Some owners are going to reallycare.
You know, if you're putting upone building a year, you care
less about the IP than if you're.
(34:53):
You know target.
You put up, you know, 50projects a year.
You see what I mean.
Like, for one group it might besomething that really matters
for their competitiveness andfor another it may not.
So I think it'll be an area fordiscussion.
Speaker 1 (35:10):
I love it, hugh, and you're always the best one to
start the discussion.
And, speaking of, you're thehost of the Constructed Futures
podcast.
We're on a weekly basis.
You host two rock stars in theconstruction industry and you're
the author of the ConstructionTechnology Handbook, which
provided a primer on technologyfor Didey's construction
professionals.
We'd love you to tell me alittle bit more about the
Construction Technology Handbook.
(35:31):
What was your inspiration andwhat did you find that helped
you transform constructionpractices along the way?
Speaker 2 (35:38):
Yeah, I think what I found was I wanted to write a
book because I wanted to divedeeper and dive deeper into the
industry and talk to the mostamazing people I could find, and
it was kind of early.
I was still pretty new to theindustry.
I'd done some things in thebackground, but not in the
middle of it, and the cool thingabout writing a book is almost
no one says no to a request fora book interview, so I got to
(36:00):
talk to some pretty amazingfolks.
I'm actually doing a secondbook, which, not surprisingly,
is called AI and Construction,so I'm interviewing a little bit
less this time.
I'm starting from a differentbasis.
The point of that book, though,and sort of the core insight of
the first one of theConstruction Technology Handbook
, was people talk over eachother too much.
(36:21):
So folks who spent 20 yearsfiguring out how to pull a
building out of the groundprobably don't spend a lot of
time wondering what API standsfor, and yet people are throwing
this word at them all the time.
So I actually spend the firstthird of the book defining terms
.
I mean, I literally define whatWi-Fi is Understanding.
Most people know what it means,but have probably never had a
(36:41):
discussion about it so thatpeople walk away saying, look, I
may have known some of that,but I know all of it now, or at
least I've heard all of it now,because I wanted to level the
playing field so that consumersof technology would be able to
make better decisions than ifthey have to kind of shut the
brain off and get through thepart with all the words they
don't know and then have to feela little stupid asking
questions because it's on themto know whether the thing
(37:05):
they're buying is what they need.
But if they're not armed withthe right terminology and
understanding of what's going on, they're really hands-strong
and that slows everything down,because now people are like,
look, maybe this would help me,but I don't understand it well
enough and I can't take a riskwith this project.
So if you give people at leastthe table stakes of what these
words mean and thinking abouthow software is built and so on,
(37:27):
they're better able to make adecision which helps both sides
it helps the contractor and ithelps the vendor of software and
it helps the people managingthe whole thing.
That was the big deal.
Speaker 1 (37:40):
So, in addition to I mean I'm really excited to hear
about this new book you have inaddition to the success of the
last one and the new book thatyou're working on, and I'm sure
it all ties together to thisquestion, which is what is your
vision for the future ofconstruction and how do you see
the work you're doingcontributing to the evolution of
the construction, bestconstruction practices and
project management?
Speaker 2 (38:01):
Yeah, I think that it's important not to get
focused on one part of what'sgoing on.
I think there's a lot of changeand there's a lot of new ideas
and a lot of new ways of doingthings that are being worked out
and tried out.
It is such an intricate as Isaid before such an
interconnected endeavor to builda building that it's really
(38:22):
hard to change one thing withoutit impacting others.
So they're all inching forward.
But I think what you're goingto see and what I, to the degree
I have a vision about this iswe start to differentiate
buildings and think about how tobuild them in the way that's
most appropriate to them.
So things like industrializedconstruction, which absolutely
relates to specifications andabsolutely relates to AI,
(38:44):
especially on the generativedesign side, but some buildings
are going to be better for thatand some are not, and apartment
building is probably better forthat, multifamily maybe less so,
or a factory definitely less so.
So you see what I mean.
Like, I think that we have atendency sometimes, when we're
talking about construction, totry to continue to lumping it
(39:05):
together, and I think you'regoing to find specializations
more and more be the case.
I think that the degree towhich a human needs to be doing.
Lots of the minutiae inconstruction, I think, is going
to go away.
If there were a vision, itwould be.
We need fewer people who aredoing more value at, more of the
(39:28):
intuition and thinking thatgoes into making a great
building, and less of the movingstuff around and less of the
paperwork and less of the stuffthat people typically hate now
anyway.
And the interesting thing,though, is often when you say
things like that, veterans inthe industry will tell you yeah,
but how are people going tolearn if they don't go through
and do 15 submittal logs?
Well, the answer is the same.
(39:51):
Technology that automates.
Some of this is fantastic atcreating little micro-trainings,
so whereas before there reallywasn't a ton of training unless
maybe you're talking aboutsafety, and even that is kind of
recent you're now able to havethe LLM or whatever the software
platform is, create littlesnippets of learning along the
way, so you're learning whatthey call learning in the flow
(40:11):
of work.
So the other thing I thinkyou're going to find is the
tools that automate and thetools that make it easier for us
to not have to do all theMimousha that, again, people
don't usually love will be thesame ones that are the source of
learning and the source ofupskilling people, not only to
where they would have been 20years ago, but, as things
(40:32):
continue to change, they're ableto be there to do what in the
learning world, we callscaffolding learning, so
supporting learning.
So you're not taking on too muchat once, you're just learning a
little piece each time, butpretty quickly you get really
good at whatever the new thingis.
So I think, rather thanautomation, necessarily meaning
that our skills go away, I thinkthere's actually opportunity to
(40:54):
be thinking and saying thesethings can also write learning.
So let's do that, let's havethem do that and deliver
learning in a way that'soptimized for whoever the
learner is.
I think you're going to see arevolution across manufacturing,
across construction, acrosspretty much every service
business, where learning in theflow of work it comes from the
(41:14):
tools you're using, notnecessarily some learning
management system that no onepays any attention to.
Speaker 1 (41:19):
Yeah, and these lessons that have been learned
are very costly to the businesstoo, right?
So if there's a way to avoidthat the cost in the learning I
think that would berevolutionary as well.
Speaker 2 (41:32):
One interesting point really quickly.
Almost every project managerI've ever talked to has a list
of lessons learned.
They often need a little bit oftranslation because they're
usually reminding themselves orthey're delivered by them to
someone else, but the point isthere are huge repositories
across the industry where peoplehave taken notes.
They've taken notes on projects, they've taken notes on specs,
on drawings, so on.
There's a lot of context outthere that these new AI tools
(41:58):
have the promise to be able todistill into lessons that make
some sense, especially when youcombine them with the work that
the International Code Counciland CSI and ASTM and some others
.
They have amazing repositoriesof information about how to
build a building and I thinkthat those things start to
become unlocked.
We're going to be able to thinkabout buildings in a much more
(42:22):
sophisticated, comprehensive waythan any one human's brain
could handle.
Speaker 1 (42:28):
So, looking into the future, as you and I just were,
I would like to transition intothe final question of the show
and the tradition of our podcast, future Construct.
If you could project yourselfout 25 years and wanted to have
any device or technology thatwould benefit you personally,
what would it be and what wouldit do?
Speaker 2 (42:45):
25 years is an awfully long time.
I want a spaceship that'll takeme to the moon, where I can
hang out.
I'm kind of coming up with thatone on the cuff.
It doesn't really relate justto construction because, frankly
, in 25 years I think the way webuild buildings will be, some
(43:05):
of it will be unrecognizable andsome of it will look a lot like
right now.
But me personally, yeah, I likethe idea of personal transport
that can I don't know about themoon, but that can take you out
into space, and I love the ideaof experiencing space and I
think I'm young enough thatthat's a reasonable expectation.
I'm not so sure about Mars.
(43:27):
I moved to Austin and say Inever want to be culled again
and I don't know that Mars isgoing to satisfy that.
Speaker 1 (43:34):
Well.
Thank you so much for your time.
It was an incredibleconversation with you, hugh.
Your journey from the 80s inHong Kong to becoming a
construction tech visionary istruly captivating.
The link specificationmanagement platform, especially
with spec GPT, seems like a gamechanger, streamlining processes
for project teams.
Your insights into the industry, challenges, your unique
(43:55):
offerings, such as yourconstructive future podcast and
the construction technologyhandbook and your upcoming book
they all provide a richperspective of looking ahead.
Your vision for theconstruction industry and the
links role in shaping the futureis insanely inspiring.
Thank you, hugh, for being athought leader and to our
listeners, stay tuned for moreinsights on the future construct
(44:16):
podcast.
Hugh, it's been an absolutepleasure having you on the show.
Thank you so much for your time.
A thanks for having me.
Absolutely Thank you.
Popular Podcasts
Las Culturistas with Matt Rogers and Bowen Yang
Ding dong! Join your culture consultants, Matt Rogers and Bowen Yang, on an unforgettable journey into the beating heart of CULTURE. Alongside sizzling special guests, they GET INTO the hottest pop-culture moments of the day and the formative cultural experiences that turned them into Culturistas. Produced by the Big Money Players Network and iHeartRadio.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.