EP 135

Paul Heming: Hello and welcome to episode 135 of the Own the Build Podcast with me Paul Heming. As always, go and check out the show notes. Today I have linked, again, the EOT template that I wrote a couple of years ago, as a subbie in the past I wrote endless EOT requests. This one’s quite an in-depth one. If you’re a main contractor, subcontractor, whoever you are, go check it out in the show notes. Feel free to give me a shout. In the studio, today we are joined by Martin Paver, who you guys will remember from episode 88. He is the CEO and founder of Projecting Success, who are a passionate team of visionaries, project managers, and data scientists on a mission to change the way we deliver projects. Martin is at the forefront of data and AI in our sector, and last time he was on the show, episode 88, go and check it out. Up until very recently was the number one most popular own the build show. And we talked kind of facetiously about the extinction of the QS, which many people took to heart and we were only half joking. But it was a really, really interesting show. I’m looking forward today to speaking with Martin kind of almost one year on, and honestly a year where AI data has probably taken quite a significant leap across many sectors. The question is, has it taken a significant leap in our sector? Martin, welcome back mate. How are you?

Martin Paver: Yeah, very good. Thanks Paul. Good to speak to you again. Thanks for getting me back on.

Paul Heming: No, absolute pleasure to have you back on the show, mate. Like I said last year was an absolute success. Just before we jump in, because there’s so many things that I want to ask you about what you are doing, what’s been happening in response, but just for clarity, for new listeners to the show who weren’t here on episode 88, who are you, what are you doing? What’s your experience in construction?

Martin Paver: So I’m Martin. I’m the chief exec of Projected Success, which is a company I founded in 2014 and I did a big pivot in 2017 and moved into data-driven project delivery. So in terms of my background, I was a chartered engineer. I dropped that a couple of years ago because I’ve not been using it. So I’m a chartered project professional and a fellow of the association for project management. And the past seven years I’ve been into data-driven project delivery. So I’ve been using that. So instead of all these lessons learned that we put into a spreadsheet and then finish it with throwing away, if we could start to leverage that data, then we can change the world and we’ll start to deliver projects with less downsides, more upside, and with greater investment certainty. So that’s the mission overall and I think we really can change the world with it, you know, exciting times so we can deliver more hospitals, et cetera. Yeah.

Paul Heming: I love that ambition. You change the world and you know, we were talking just before we hit record, didn’t we before we came on air, that in the world of construction where everybody listening is probably, I’m going to sound negative here, but everyone listening is probably on a job that is either over budget or in delay. It’s one of the two. I’d put good money on that fact and just, we all love our sector. We all love the projects that we’re delivering. There’s so much opportunity for change, for efficiency gains, for productivity gains, whatever you want to call it, that we can literally change the world with innovation. I feel I’m at the heart of that was what I’m trying to do. And I know from speaking to you in the past that absolutely it’s the same. And I remember talking to you when we first sat down and you were talking to me about you were going over lessons learned and you were kind of doing it in this old fashioned way. You know, we’ve all been sat in those meetings, haven’t we? Where it’s right. What did we learn from this project? Get out a Word document, get out of Excel document or whatever, go through it, have a chat about it, have a good conversation about it, and then kind of toss it in the bin and move on. As opposed to the new world which we’re trying to signal, which is out there, right? AI data, machine learning around actually creating machines to do this for us. Just go back to the thinking around that because I think it really helps to ground the conversation. Why did you all of a sudden think, what am I doing with this old school lessons learned mentality?

Martin Paver: So I pulled together 20,000 of these lessons learned, right? I got a really big data set and I made myself very unpopular by putting in FOI requests, right? But it’s something I felt in terms of society we’re just throwing that data away. We’re not leveraging it. And I felt that we’d got a responsibility to change that. So that’s what I’ve been driving for the past, I don’t know, seven or eight years, something like that. I’ve been really, really keen to change the view on it. And why is that? So if you look at that data, it’s very sort of one dimensional. It’ll probably tell you what the project is and what the lesson was. It doesn’t necessarily tell you the impact of that lesson. So you can’t prioritize. It doesn’t tell you if it’s avoidable or not avoidable, right? So it might be an act of God, it doesn’t tell you about context. So you can’t sort of replicate it in your current instance and it doesn’t tell you about, did you try it and head it off at the pass? If so, was that successful or not? So all of that richness in that data set is gone. We’ve lost it all and that’s because we throw it away. So we boil it up into a spreadsheet and stick it in probably one or two cells in a spreadsheet. So if you just imagine the minority report, did you see that Paul? Did you see that film?

Paul Heming: Long, long time ago. So don’t ask me to go through it in great detail, but yes.

Martin Paver: So just imagine that you could sort of fast forward into this pre-crime, right? You could predict when something’s going to go wrong. So just imagine, go backwards and forwards, backwards and forwards, backwards and forwards on that same event and then see when it trips over. So if we could do that with a project and you can fast forward and see which parts of a project then became sort of difficult and why, that’s what we’re trying to replicate. What we’re also trying to do is a bit like a sat nav. So if I know is my school is kicking out at three o’clock and it’s going to cost me a 20 bit delay, I’d go a different route. So that’s what we’re trying to do with a project is that if a certain course of action at a certain time is going to cost me something, then I’ll try and find a way around it. So if I can codify my dataset such that I can start to get that from my dataset, then great, I can start to get all these insights and that’s exactly where I want to be. And I think that’s the challenge.

Paul Heming: I remember, it’s all coming back to me now Martin, and it’s almost makes me chuckle a little bit because I remember you were making freedom of data requests on all of these public projects essentially that should be in the public’s gift and the public’s ownership, right? We’ve paid all this money for hospitals, schools, railway, whatever. And you were saying give us this data and there was a lot of pushback because people don’t want to give away data, but that data kind of then comes together to create a baseline and then from there you can do your AI, your machine learning and try and see the trends, et cetera. I’m no AI machine learning expert, but I understand that at the core of it you’ve got to have a really rich, really strong data set to have any chance of being able to learn. What I feel from the outside looking in, and I’m sure you’re going to tell me that I’m a fool here and feel free to do so, is that since we spoke in the summer of 2022, late summer of 2022, I think it was quite a lot has shifted. There was a lot of, I remember speaking to you and you felt there was a lot of skepticism, AI, machine learning, data construction could never happen, yada, yada, yada. And then along comes chat GPT and people’s mentalities on a mainstream level, I feel has shifted as they’ve started to become more aware and actually use some of these tools. Right? My question to you is from that, you seem chomping at the bit to answer it, but from that skepticism or where we were in 2022 when we were chatting on episode 88, how far forward have things shifted in your view for construction?

Martin Paver: It’s quite split actually. I think it’s now sort of mainstream. So it’s now on lots of people’s attention. I think the boards can now see the scale of the opportunity. They can see that there’s jobs you can start to do differently. So one example is, so I mentioned previously about the training that we run. It’s all paid for through government, so it’s free for a lot of organizations. And I spoke to a marketing director for a big construction company and I just showed them a mid-journey. So that’s the image generated tool and it produces some phenomenal art. And I showed them Chat GPT, and they said, right, so just tell me about your topping out Sony and the sort of copy that you’re looking for. So they told me about that and then they said, right, so what sort of image would you like from your stock photo? So they said I want sunset, but it’s a sunset with a rainbow over the top of the shard. And that’s exactly what I gave him. That picture is exactly what I gave them. So that person has said to me, you’ve just destroyed the entire team that I’ve got. You’ve start to reimagine the way we’re going to work. This is phenomenal. So I’d like to put some people through the training and work out what this means. So that’s exactly what we’ve done. We’ve put them through the training, they’re now start to reimagine the way they work. So in terms of all this social media, so what you can do now, you can start to sort of write something and say point that text at an 18 year old female who’s about to join construction and make it relevant to them. Make it relevant to a 55 year old bloke who’s looking for a fifth career or something like that. So you can now start to take the same message and now start to hyper-personalized it to different segments of your population. So you then—

Paul Heming: 10% of the time.

Martin Paver: A fraction for the time and you can start to do all the AB testing, see which messages are sticking, which ones aren’t sticking. It’s a different sort of job. You’re not just writing a load of words and sending out social media as you’re now sending out all this hyper-personalized content, which is all sort of targeted around people for different reasons.

Paul Heming: Yeah. I can see with the copywriting, content writing, all of those things we do in our business where these models can effectively do 80% of the work for you or the grunt work. If that’s the right way of putting in, then you can finesse it all kind of as you were suggesting, do that. I guess what does that mean, there’ll be QSs natural cynics listening to this thinking, yeah, but you couldn’t do my job in the same way that you were. We were talking about it last year and I got a barrage of people saying, no, you can never do that, you can never do that. And we didn’t act, I know we joked that it was the extinction of the QS, but really it’s the evolution of the QSs and that’s exactly what you’re talking about there. It’s re-imagining people’s roles. So contextualizing it with those project professionals. What shift do you think that chat GPT as the obvious standout model has had on project professionals?

Martin Paver: So I think the challenge at the moment is security, right? So there’s some organizations who’s locking this down and saying you can’t use it. And that’s because there’s been some big international companies who’s used it and somebody’s copied in a load of sensitive data and that’s gone back in its training models. So if you ask the right question of that model, it’ll spit out somebody else’s confidential data, right?

Paul Heming: Because it scrapes the internet effectively, right?

Martin Paver: It’s because if you put some data in and say, summarize that confidential document for me, it will do that. And that goes into its training model. So if somebody else asks a very specific question about a specific piece of technology and that happens to be your technology, then it’s going to reference back to your technology. So that’s the way it’s doing it. So there’s some organizations who’s actually turned this off. So they said you can’t use chat GPT. So government’s come out with some policy on it recently, I think it was last week or the week before and it said use it, but you’ve got to be educated, right? You’ve got to be educated about the downside risks and make sure you use it sensibly. And where we’re going to get to is these large language models of which chat GPT is one example. There’s a load of them which are open source, so you can now start to run them locally on your laptop. So it’s like its being trained to speak English and locally you can give it a dialect, which is all your business language, all your rules and all that sort of stuff. And that doesn’t leak back to the English language, it stays in your local dialect.

Paul Heming: To the C language or the projecting success language or whatever is local to your diet. I guess. But it goes back to surely right and maybe I’m jumping ahead here, but the problem that you were telling me about 12 months ago was a lack and going back to your problem where you had these 20 thousand lessons learned with data all in a different structure, is the problem not with construction that yes, you can say give me some content for a topping out ceremony, because that that’s all over the internet, it’s press releases, whatever. But with construction, our data company to company is generally held in secret or in silos. Like it’s not shared. So is it actually something that can drive us forward or do we have that same data problem in the short term?

Martin Paver: So if you are Belford Beatty, right, you’ve Britain’s biggest construction company, they don’t want to pull their data because they’ve got the biggest training dataset. So they’re going to win. So why would they want to open up their data? Now if you are number five, six, and seven and you get together, then you’ve got the far bigger dataset between you plus as well you’ve got probably a greater breadth and diversity of projects. So if you’ve got that, then you’d get a better and better training dataset? So that’s what we’re trying to do is we’re trying to start off on things like health and safety. So on the news today, the construction news website said 45 people have been killed in construction in 2022 to 23. Now that’s the biggest number for a long time. And that’s just the headline number. All the people who’s been injured and it’s been life-changing, all the people with long-term sickness, et cetera, we’re not pooling the data which underpins that. So a few months ago I was speaking to the ONS, so the Office for National Statistics and they said to me, we can’t connect the long-term sickness data with the causality. If we could start to get a link between those two data sets, then we can start to understand what we need to do to drive down that risk. What we also need to do is to understand the correlation between offsite manufacture, on site manufacture. What does that mean to delay, what does it mean to carbon? What does it mean to health and safety risk, et cetera? All of that’s in our data set if we leverage it, but we don’t. And people see that that’s got a load of tactical advantage if they keep their own data back. And this is where the clients come in. The clients need to be mandating this stuff.

Paul Heming: Even if the client is the government and the client can mandate it.

Martin Paver: So National Highways, so a chap called Adam Perkins is working at National Highways and what he’s looking at doing in their next risk framework, so their five year framework is to say, I’m going to reward people in terms of give you more scores if you are working collegially. So what does that mean? Does it mean that you’re pooling your data for the greater good? So if you’re building one road and you’re building the next road and the next road, are you pouring your data into this? So as a client I can build better and better roads. So that’s what I want as a client. Am I making sure all my people go home safely at night in terms of solutions that we’re creating? So if we create a product which is basically a configuration of some of these open source tools I’ve been talking about or maybe the Microsoft stack, if you are pooling that and you can drive it down your supply chain all the way down your supply chain, it means you get data standardization. And a lot of these main contractors, they tend to compete at the moment on the basis of main contractor performance, but a lot of them just are bolting together, lots of subcontractors. So a lot of your data actually sits in the supply chain and their fire to the moment on their own data. And that’s just a fraction of this overall data. So like the government data, I’m finding that they’ve probably only got a maximum of 5% of the total supply chain data. So when government’s saying, well, we’ve got all this data we can use—

Paul Heming: Because the main contractors hold it and then yeah.

Martin Paver: Yeah, it’s like HS-2, right? There’s no big central bucket of HS-2 data, it’s all stuck in the supply chain. So all this problem we’ve had getting hold across rail data, we’re going to have it all again with HS-2 data. So if we want to learn lessons from all of that HS-2 data, it’s all there.

Paul Heming: Yeah. No, I’ve got a good friend who is quite senior in the commercial team at HS-2 and just the fragmentation I’ve heard about and you just think really on a project like that, we can’t even get that right. But we probably shouldn’t go into that and start talking about a different project. It’s probably not a wise place to go. So in short, it feels to me like you’re saying, yes, things have shifted forward, but no, we are not in a transformational new world 12 months on and the world hasn’t changed in terms of that key fundamental is breaking down those gates if you like, to the data so that every project can benefit from it as opposed to just those companies who’ve got the big data sets. Is that fair?

Martin Paver: I think it’s bigger that Paul, I think it’s now on the C-suite radar and it wasn’t before, it had to fight with COVID, it had to fight with productivity and drones and Bim and everything else. And I think it’s a lot higher up on their prior list now because they can see the scale of the opportunity and they can see some of their competitors moving. So I’ll give you an example for instance. So a couple of weeks ago I was invited to go to Belfast for the government heads of project delivery profession. So it was a one and a half day session on project data on ethics. So they invited me along to do an after dinner talk for 45 minute slot and he went on for over a one and a half hours. And it’s people like—

Paul Heming: I’m assume the volume of questions, not because you were on a wild rant telling them…

Martin Paver: They both.

Paul Heming: Talk a bit of.

Martin Paver: Both. No, no. So, lots of questions are coming by and they’re basically saying, so why aren’t we doing this now? A lot of them were asking, so what’s stopping us? What’s getting in the way? And there’s nothing getting in the way apart from, its things like intelligent customers, right? Have we got enough intelligent clients out there who can ask the right questions from an understanding of this technology? If we’ve got that, then we can start to shape it. If you start to put it in contracts, then you can start to shape it. If you understand what the data model looks like, then you can ask for the right data at the right time. And I think part of the problem is, is project people are busy doing projects, they’ve not got the head space, they haven’t got the time to be able to drive this forward. So it then falls to the data people and the data people don’t necessarily understand their domain. So what you finish up doing is squeezing more and more performance out of the algorithms, but the input data, it just crappy data. And that’s part of the problem is that if the data’s not good enough…

Paul Heming: Is that why you need QSs who do understand the domain, who do understand the challenge to start retraining, to be able to convert it into meaningful data sets and the like?

Martin Paver: Absolutely. So last time I talked about James Garner, right, phone gleeds is come on leaps and bounds. He’s just got his data IQ a hundred. So one of the most hundred, most influential people in data he got that last year or this year, sorry, and is now taking over my slot. So I stood down as the chair of the project data under its task force in 2022, the end of 2022, I’d done my shift and James has now taken that on. So a QS is now leading a lot of this and is now interfacing with government is going to be joining the APMs data advisory group, et cetera. So he is a man who’s probably had quite a significant career shift and the world’s his oyster.

Paul Heming: Yeah, 100%. And you can see, I know James and I’m actually saw him at digital construction week and you know how much he believes in this. But actually we’ll take a break now and we’ll come back and I just want to talk to you a little bit then about the project brain that you’ve been talking about and also someone like James and that journey that you can go on with your apprenticeships, et cetera. But we’ll come back and we’ll do that right after this break.
So you know, I love a QS and I know that you are not a QS but you’re going to have to be nice to QSs Martin for we’re important, we’re important. So we touched on it there at the end of that show and in the last show, in the show notes we left details of the apprenticeship, we’ll left details of the hackathons. You do just quickly explain before we get into the heart of today’s show, what you can do if you are a QS listening to this, who thinks I see the future in AI in data and I want to retrain alongside what I’m currently doing, how can you help them?

Martin Paver: So if they’re based in England, then there’s a load of money that’s sat there in the apprenticeship training levy. So just see it as a tax on government. They tax pay bill of more than 2 million pounds at 0.5%. So that sits in this training account and that money is sat there to make UK productivity better. So we’re supposed to be upskilling people to make productivity better. So it’s there, it’s available for you now, it’s been there since 2017 and if you don’t spend it, then two years later it gets taken off you by the tax fund and they keep it. So it’s free training money and most organizations aren’t spending it. So what we offer is a level four apprenticeship. So that’s a bit like a foundation degree and that gets you going. So it gets you understanding the principles around sort of artificial intelligence about data models, ontologies, data security and the way you can fit data together and graph databases and things like that. So it just gets you going and that means that you can have a conversation then with your data team and your project delivery people and broker between the two. So we call that a translator function. So that’s the course that James Garner did. So we talk about him in the first half. So James did that and he got the best score that we’ve had so far. So he got a distinction and he did really well in it. So what we’re just about to release in August is a level four course, but it’s going to be niched around risk management. So risk management is the human process of guessing what’s going to go wrong. If instead we can now start to use the data, then we can use the data to drive a different approach. If there’s enough demand, then I can put one on just for QSs. So project data analytics four QSs and we’ll bring 25 people together and we’ll transform the profession. So I can run that for you. If there’s enough demand out there, we’ll run a niche cohort and then Paul, you can have these people on your show and they can talk about all the cool stuff that they’re going to produce, which will change the industry, right? And we’ll open source the lot and then we can use that to drive that transformational change.

Paul Heming: And we can change the world and the world will be a better place, right?

Martin Paver: Yeah, absolutely. This is more sort of energy transition projects, right? We’re building 40 hospitals in the UK. We need to be taking all of that learning from each hospital and feeding it through to the next one in terms of performance curves and productivity metrics and sort of technical queries and all that sort of stuff. All that needs to be smashed together and we get better and better and better at it. So it takes a QS job and then turbocharges it, that’s what you’re trying to do.

Paul Heming: Yeah. Completely understand that. I completely see that. And so layman’s terms, there is a load of government money in this apprenticeship levy scheme, which if you are a QS, young or old and you think that AI data, all of these things are things that you want to be involved in, you believe in, you see the long-term vision in, what you can do is feature your employer who can get access to this levy. And in effect you are going to go to university and inverted commas for free to safeguard the future of your profession. Is that in one word? Am I right? Yes or no?

Martin Paver: Absolutely. Spot on. It’s a no brainer, isn’t it? It’s a no brainer.

Paul Heming: Yeah, I mean if you’re listening and I’ve spoken to, since episode 88, I’ve spoken to so many people in the last year who get it. So distilling it like that, there is no reason for you not to do it. The money is there, you’ll be paid to advance your knowledge and understanding of that part. And I totally agree with you, Martin, it is a no-brainer.

Martin Paver: It’s not just about the profession, this is a risk mitigation against your career. So if this stuff really, really takes off and it starts to threaten your career, if you are part of the vanguard, you are now mitigating against that risk. So if I’d got a load of bills to pay, if I’d got a big mortgage and I’d got kids at school and all that sort of stuff, I’d be thinking, why wouldn’t I do this? What’s the reason that I wouldn’t do it?

Paul Heming: Yeah. And the point is construction is a laggard in terms of its technology uptake, right? So there is still time to do this and if you’re listening, I really recommend that you do take a look at that. Getting into what we’re talking about now Martin, go on. You want to say something? Hit me.

Martin Paver: So just one more thing is that we’re just about to launch in September as well, a level seven course. So that’s a master’s course. It’s going to be artificial intelligence for project professionals, right? So it’s a standard syllabus around artificial intelligence, but it’s aimed at the project community. And there’s some people who thinks, well, I’ve done a degree in civil engineering so the next thing I need to do is a master’s. And I would say just be careful about that. So we can learn a lot about artificial intelligence, but if your data is not good enough, you can spend all your time tweaking algorithms, but the source data you’re pushing into it will mean that you won’t get good results out of it. So that’s why we need to focus on things like data models and data pipelines, pulling sort of information out of spreadsheets and pulling it out of the emails and pulling it out of online systems and web scraping and all that sort of stuff. So something we did with gleeds, for instance, in a hack was a simple web scraping. So you can go to the screw fix website for instance. So in simple terms, go to the screw fix website and I can download the price of screws today and tomorrow and the day after and the day after and the day after. So if you think about the inflation metrics that we use today, I don’t need to use those anymore because I’ve got component pricing on every single part of a construction project in real time.

Paul Heming: Yeah, there’s so many things, isn’t there? Just even that on the tiniest level tells you the opportunity, doesn’t it? That you could get real time pricing on all different kinds of asset classes constantly, as opposed to what going back to BCIS or going back to wherever and trying to work out what it is and then indexing it forward, et cetera, et cetera. It’s nonsense. The more I think about it, the more I think it pains me now that you say that we haven’t kind of jumped as far ahead as I would’ve hoped we would’ve done in the past year. But therein lies the opportunity in some respects, right? So you recently said Martin, and I’m going to quote you here, if we aspire to make the most of the advances in data science and AI, we need a project brain, therefore we are building one end quote. Could you tell me what you mean by that knockout sentence?

Martin Paver: A question back to you, right?

Paul Heming: Oh God, don’t do it.

Martin Paver: If you are trying to build a data model for a project, then where would you start?

Paul Heming: Well, I’m a QS so I would probably start with trying to understand what you’re building and how much of it there is and how much it’s going to cost.

Martin Paver: Yeah. That’s the place everybody starts because that’s the way we’ve been trained.

Paul Heming: You’re going to tell me I’m a numpty now.

Martin Paver: No, no. If you’re trying to solve productivity, right? We start off with the KPIs that we’ve been using for years and years and years. So we’d look at waste, we’d look at gross value added, we’d look at those sort of metrics, but it’s a dead end. We’ve been looking at that data with the government construction productivity task force, sorry, and it’s a dead end so you can get your hands on all of that KPI. So if I’ve got loads and loads of stats on gross value added, I don’t understand the causality and the differentiation between sort of one contract and the next contract and I’ve just got the top level number, I can’t drill into it because that’s all I’ve got. It’s just the KPIs of the metrics. So if we flip that problem upside down and we say what we want to do is we want to solve the project delivery problem statement, right? It’s a logic puzzle. So just imagine why do we do risk management? We don’t do risk management to manage risk, right? We do risk management to manage the downside, to minimize the downside and maximize the upside to the lens of different stakeholders, right? So that’s what we’re trying to do with it. That is not just risk management, that’s things like change control logistics and things like that. So if we can now start to get the data feeds in place, I can say these are the problems that I’m trying to sort of manage. I’m going to accept that there’s change control. So there will be client changes. But if I can pull those in at the right time and say to my clients, once you’ve gone past this date here, I’ve seen that you are predisposed to changing your mind later, if you do that, it’s going to push your price up. So this is the decision point and this is the evidence why if you choose to override that, then that is your choice. Alright?

Paul Heming: Just stopping there you can actually see that, can’t you? If you had a data set which was rich and the client is thinking of changing, I don’t know, a feature on the project and you could kind of say, look, at day X it’s likely to cost you 100 pounds, but if you go forward through today, why all of our data says this is going to cost you 18 weeks and 200 pounds. Are you sure you want to do it? And they then go, let’s do it. As opposed to being like, wow, that I did not, I appreciate the context and gravity of that decision. They could then change their decision making, right? But we just do it blindly, don’t we?

Martin Paver: Yeah, we do, we do. So if we can start to spin all these different permutations as well and look at all these scenarios. So we’ve got generative design today, so why don’t we have generative air carbon models and generative productivity models and generative cost models where it optimizes for each of those parameters. So that’s what we’re trying to do ultimately. So we can do it for design. So why don’t we do it for all these other parameters? Because it’s still a logic model. So if we take this risk management problem and redefine it and then break it down into these problem statements, that’s what we’re trying to do basically is to solve those problem statements. So just imagine it’s—

Paul Heming: That is the project brain.

Martin Paver: Well, it’s a logic puzzle, right? So if you see that that is a logic puzzle, you’re trying to solve each of those problem statements one by one. So if we take a problem statement and we say, have we got the data to solve that problem statement? If you haven’t, you need to get the data from somewhere, right? So you need a data solution that pulls the data out of somewhere. It might be an app, it might be a script that rips it out of an email, it might pull it out an invoice, it might web scrape whatever, right? So it takes the data. So you’ve then got to pull the data together and integrate the data and you need a second data solution. So that’s your input data sorted out and now your output data, it needs to go into a tool which will give you some insights. So you can answer that problem statement.

Paul Heming: But can I stop you there and just ask, because it’s for me starting to get a little bit abstract. Now I really, really like to contextualize this so I can understand it. So as if for instance, what’s a problem statement that we could either, that you could say is an obvious problem statement on a project?

Martin Paver: I want to minimize waste.

Paul Heming: Okay, so problem’s name is I want to minimize waste. Have we got the data was your next point, right?

Martin Paver: Yeah. So what I’d do is I would decompose that and say, right, so there’s seven different sorts of waste. So I’d break all that down in terms of different sorts of waste and then I’d say, right, I want to understand what’s going in the skip and the stuff that’s going in the skip. I want to be able to classify it. So does that mean I’ve got a camera on top of the skip so every time something’s thrown in and I’ve got some weighing scales on the bottom of the skip so I can classify it and I say that’s a piece of wood that’s going in there and it weighs this amount, right? So I can now start to solve that specific problem statement around what is going in my skip. I then say, well I want to avoid that in the first place. So can I start to order timber which is pre-cut? So it’s pre-cut in the factory and it comes along as a big sort of Lego pack and it’s got numbers on it and I can just use it as per the job. I can drive down my waste because that waste is then created in the factory and it’s reduced because it’s all optimized based around the length of wood that you’ve got. So it’s then saying, what can I do in terms of that overall ecosystem so I can minimize this problem about the amount of wood going in a skip, right? That’s a very, very specific and small use case.

Paul Heming: Yeah. No, but it helps you to understand it in a real simple terms, right? So then if we zoom out, so company main contractor says, I want to reduce waste and then they’ve got their seven categories of waste or whatever, right? The main problem, and it’s a really nice simplified example is that what waste are we producing, right? So it’s almost like to start with we don’t know that. Once we know that, then what do we do with it? Is that what you said?

Martin Paver: Yeah. And is that waste because somebody’s made a mistake? So is that a quality issue or is that just an off cut? So where’s it coming from? Is it a byproduct of a rubbish process or is it a byproduct? Just the way we do it today is that we always order bits of timber that’s too long for the job. And so once we start to package that data up, if you’re solving that one problem statement, what that’s also doing for you is it starts to look at things like tool time. So that’s another form of waste is that if you’re not on the tools, you’re sat around waiting. So what you’re trying to do is to optimize tool time, right? You don’t want people on the tools all the time because that’s not realistic. They’ve got to get from A to B, they’ve got pick materials, et cetera. So you’re trying to optimize around sort tool time. So what does that mean in terms of another sort of waste and is the connection between that and getting the right size of timber. So if you’ve got the right size of the timber all the time, bang, bang, bang and you can just go and put all your plaster board up, it won’t go, then that’s going to increase your flow and it’s going to increase your overall productivity. So there’s a connection now between the waste that goes in the skip and the waste in terms of tool time and the loss of that productivity. So that’s the connections in the dataset, which is a data model. So I can join personal productivity with the training that you’ve had, the skills that you’ve been given, the experience that you’ve got, et cetera. I can join all that together and I’ve now got a data model that is called an ontology, right? So it’s the way your data joins together is the way that you optimize to solve the problems. So if you can solve all of these problems, you’ve now got the data feeds in place to solve each of these problems one by one. And once you’ve got the data feeds and you can solve your problems, it doesn’t mean you’re solving it in the best way. It just means you can now answer it. So once we’ve got all the data pipelines joined up with the problem statements, it’s now an optimization problem. So that’s where the PhDs start to come in and say, right, if we do this, this, and this and join this up and dry that and lots of scenarios as well. So if you can get scenario generation on it, then you can fully optimize for whatever you want to optimize against carbon. Is it against waste, is it against productivity?

Paul Heming: No. It makes sense. So let’s go back to the original question where you said, I can’t remember exactly how you phrased it, but you said at the outset of a project, what are the data points that you would need? So given what you’ve just said about the problem statement and the project brain, like how are you saying you should be, what data points should you be using at the outset of a project as opposed to my typical answer of I’d want to know what we’re building, how we’re building it, then I’d measure it and I’d price it. What’s different?

Martin Paver: So I think the challenges at the moment, if you take an organization such as National Highways, it wants to put the construction data trust into its next risk framework, right? Or something along those lines, it’s going to say what are you doing? So I can learn from one project and roll it into the next one even if I change my supply chain. Alright? So that’s why the construction data trust is there, so we can securely pull that data. So if you don’t ask the right question at the start and you say, right, just give me the data, they’ll say, so what data do you want? And they say, well I don’t know. I want these six fields. You say, is that all the fields you want. Do you want those separated? Do you want them every minute, every hour? You know, what do you want and why? And nobody solved that problem, right? So that’s the concept of project brain is that we take a thin slice of the brain and say let’s solve that problem for productivity, let’s solve for health and safety, let’s solve it for carbon, let’s solve it for something else. If we’ve then solved that logic puzzle and we’ve got the data feeds to solve those problems, then we can open source….

Paul Heming: Any project could feed into it.

Martin Paver: Absolutely bang on. Right? So just imagine as we put that project brain into the data trust and it sits in the data trust as an open source product. So that is basically, it’s something I call baby brain, right? So when you’re born, your brain’s all wide up, you’re just not seeing anything yet because you’ve just been born, right? So there’s nothing in it. So it’s wide up and that’s what the ontology is, that’s what the data model is. It’s saying this is the way that your schedule should connect to your waste, should connect to your skills, whatever, right? And then what you start to do, you start to overlay on top of that in the data trust as McAlpine’s data, Macy’s data, Alpha-beta’s data, loads and loads of data over the top and see which of those data sets are most aligned with the problem statements because those are the organizations who are flying the plane with the best insights and there’s some organizations who’s flying the plane with their eyes closed in the fog in the dark.

Paul Heming: My baby brain is just about starting to comprehend and appreciate what you’re saying here because, so are you building the framework? Is the project brain going back to that quote of yours, building the framework for anyone to come in and be able to submit data so that you have a fresh data set that is logical and aligned across project.

Martin Paver: So there’s two parts to it. The first one is we’re going to open source the data model or the ontology, right? So that is a connection between the problem statements and the data. We’re then going to open source some of those tools that create the data in the first place. So it might rip it out somewhere and then we’re going to open source those tools which gives you the insights. So it’s the problem statement through to the data with the data generation and the data insight tool. So if we’ve got that, we’ve got the end to end and that will be available to everybody. Alright? It then becomes a data volume issue before it starts to drive artificial intelligences that you need the volume of data plus the connections in the data so you can drive that. So that’s where the data trust comes in. Because you can now get the volume of data once overlaid to project brain. You’ve then got the volume of data in a structured way that brings consistency. So that’s the aspiration.

Paul Heming: Yeah. No, I can understand it and we’re going on and on here, but it’s a very interesting topic. My final question for you, and this is going to be a pretty difficult question to answer I think, but if you consider the journey from no AI and data being used on projects as the start point and this utopia that you have in your mind of whenever it happens that all projects are feeding in the same data, it’s being used, it’s being analyzed as machine learning applied to it. It is what you picture construction could be where you change the world. At what point are we in that journey from naught to a hundred? Do you think we’re at 10? Do you think we’re at 50? Do you think we’re at one? Where would you place us?

Martin Paver: Somewhere between one and three, right? I think we’ve got so much further to go. I think that’s part of the problem is that people say what we need to do is go and buy a silver bullet, right? Go and buy chat GPT or something like that. The problem is the data is not in the right format. It’s not good enough today. It’s to be able to train these big algorithms that’s not consistent, right? So there are cases and plans doing some cool stuff. There’s other people doing some cool stuff, but it’s only for a niche point solution from all of that problem statement. And what we’re trying to do is get all the data joined together so you can solve any problem because you’ve now got the ontology to do it, right? You’ve got the way the data all joins together. If you’ve got that and it’s appropriately structured, then you can throw students and data scientists at it all day long and they’ll come up with insight after insight about optimization. And where we’ll eventually be is you’ll press a button and it’ll optimize against all these parameters we talked about, carbon offsite, manufacture on site, manufacture productivity. Is it windy, so the cranes winded off or not, et cetera? So we did that at a previous hack. We’re saying if there’s an extreme weather event coming in six days, what would you do to reschedule your project? We have already solved that problem and we opened sourced it. So we did that HS-2 as a few teams worked on it. So if we’ve got that, that’s now a module of capability that people can pick up and start to use it. So if you’re now optimizing because of extreme weather events, then why can’t you optimize for everything else?

Paul Heming: Yeah, well, I mean there’s so much that you can do. And just taking it back to what you are saying about the apprenticeship levy, right? You are probably one of the biggest experts in this field in the UK construction sector. You’ve been passionate about it for a decade or so now, right? You are telling us that we are one to 3% of the way along our a hundred step journey and you are saying we are, the government wants you to retrain or train into these things and we’re going to pay you to do it. You know, if you’re a project professional who thinks, who believes in the AI and data and what it can do for the future of the profession, why on earth would you not take that chance to go and not retrain just side train on these other steps? It makes perfect sense to me. I have to say, Martin, your ongoing enthusiasm and energy and passion for this topic is fantastic and I can see that you know these, it might feel like baby steps, but it’s not. It is making significant change. Its break but breaking down those barriers, isn’t it? To long project getting into the next project and actually seeing data start to be sourced the way we want it to be, right?

Martin Paver: It’s a massive frustration, Paul, is that I can see this as clear as day. It’s just there in front of me. I can see it all. And it’s almost like the matrix, it’s like…

Paul Heming: No one else can it.

Martin Paver: The matrix, I swallowed the red pole and it’s all blatantly obvious to me. There are some others who’ve seen it with me like James Garner and Grant Fendi from the Caliper, right? Some of these others can see it, but it’s probably 10 people, maybe 20 at a push. It should be thousands and thousands of us. If we’ve got thousands of us, we will drive the acceleration of this. UK will be a world leader in it. You know, government’s behind it. It’ll be in future contracts. There’s nothing stopping us apart from a horsepower, right? It just needs a load more people. And if you get upskilled in this, it’s like saying, so back in the day when Excel came out, right? And I can remember those days, I don’t do like spreadsheets and stuff because I’m not a finance person. So let’s somebody else do Excel and I’ll just do what I’m doing. It’s exactly the same argument today. You do not have a separate Excel team. Yeah. So what…

Paul Heming: I’ll stick with the BOQ on paper actually guys, you crack on with it on Excel.

Martin Paver: Yeah. So that’s exactly where we are today. There’s just a new generation with this stuff. You got to keep up with it and the difference is as well. So Excel would enable you to work more efficiently and you can do the maths a bit easier and it’s more systematic with this stuff. It enables you to reimagine your job. So I give an example is this is not just back in Henry Ford State, you know if you go to your clients and say what do you want? They want faster horses. So nobody’s seen a car, they don’t know what a car looks like. What this does, it enables you to start to envision what a car looks like so you don’t just automate what you’ve got today. Because that would be a really bad thing if that’s all we did because we’re automating a crack process. So we need to optimize that process now based on all this technology that’s coming out. So it’s not faster horses.

Paul Heming: And guys, it has been almost a year, like I said, since Martin was last on the show, it’s almost 50 episodes ago. We will probably do another show in 50 episodes time and then we want to have our first cohort of own the builders who have gone through and started to change the world because honestly it is abundantly clear what you could do and where the journey, where the technology is taking us. And if it was me and it isn’t me, I’m not in the QS world day to day now, I would be doing this because it makes perfect rational sense to me. I will leave in the show notes Martin’s details, projecting successes details and I’m going to ask Martin to share the details as well of the apprenticeship where you can get hold of that. Martin, thank you so much for bringing your energy and passion to own the build for a second time, my friend.

Martin Paver: It’s my pleasure Paul. And I’ve got a guarantee for you as well. So if your listeners out there want to get involved in this cohort, right? So I can put one on which is own the builders. So if we get 25 people, we can work with you and we’ll work with you and say what problems do we want to solve through that cohort? So we’ll train them how to do it and there’s no exams anymore. It’s just about them working on individual projects. So they’ve got to work on three projects each as part of the course. In total that’s 75 projects against three projects each, 25 people. If we do that, just imagine 75 projects worth of insights. That’s when you’ll move to dial. And it’ll be own the builders who’s moved this dial. So it’s the power of community and the power of the network.

Paul Heming: We’ve got to do it. Yeah, well there’s thousands of people listening so I’m sure we can do it. So let’s, that is the challenge that has been laid down. I’ll put the details…

Martin Paver: Let’s make it up, Paul.

Paul Heming: In the show notes and let’s see it happen. Martin, thanks so much for coming the show…

Martin Paver: My pleasure, Sir.

Paul Heming: Guys, I will speak to you all next week. Cheers.