Data-Driven Construction Project Controls Using 4D, 5D, and Predictive Analytics

NIRAN SHRESTHA: Good morning, everyone.

AUDIENCE: Good morning.

NIRAN SHRESTHA: Did you guys have fun last night?

AUDIENCE: Yeah, we did.

NIRAN SHRESTHA: Always, every Wednesday night, in areas like that-- So, thanks for coming over. I'm sure you guys had a great time, and then now had enough coffee to keep you awake. And very coincedentally, you know, we are also talking about something very interesting. So hopefully, we'll be able to keep you awake for the next one hour. And you can feel free to ask me questions along the way, because otherwise, it gets really boring.

So I'm Niran. I'm the CEO and co-founder of onTarget. onTarget's a visual scheduling software with predictive analytics. We are [INAUDIBLE] partner. We've been using Forge as our backbone, and building software on top of that-- based out of New York, a two-year-old company.

Just for my background-- I'm sure you've seen it on the website. But my background is architecture. And then I went to do some facilities management. And then I went into construction management, civil engineering, and then scheduler. So A-O-C-E-- that was the sequence of 10 years of my life.

And I've worked in Africa for around four years, and New York for around seven years, now. I'm originally from Nepal, so a couple of years in Nepal. So I worked in some of the highest mountains in the Himalayas, as well as in Ivory Coast, where I there was a war, where I was building bridges for rehabilitation after the war. And in New York, I worked in Second Avenue Subway, the east-side access, and the UN Headquarters building. So those experiences led me to the realization of the potential of the data in construction industry.

I went to school at Columbia University, and did my research in data analytics. So my co-speaker is Max.

MAXIMILIAN SCHUTZ: Good morning, everybody.

AUDIENCE: Good morning.

MAXIMILIAN SCHUTZ: Yeah, I'm Maximilian Schutz. I'm a visiting scholar at Stanford University right now. I've been doing research together with Martin Fisher for the last year and a half.

Before that, I worked as a implementation consultant for [INAUDIBLE] and another GC in Germany called [INAUDIBLE] one of the largest GCs over there. Before that, I started out as a concrete worker in a dual-studies program, bachelor's degree, at [INAUDIBLE]. So I've been there for 10 years now. I did my master's together with Martin at [INAUDIBLE], so it's always back and forth between Germany and the US, kind of, for me. And now it's the US time again, and we're doing research together in this area as well.

And Niran called me up. We published a video on our vision of feedback loops in construction planning. And as it turns out, our visions aligned very well. We have the same understanding. And we thought, why not give a talk together? And it's a really great synergy together, to present our topics in that way. So I hope you enjoy it. And I'll hand it over to Niran again.

NIRAN SHRESTHA: Yeah. So the sequence is going to be that I'm going to give a brief introduction of the-- not introduction, but what data and how the analytics works. He's going to go through his slides. And then we're going to finally give the solution of how do we solve the problem of data in construction, actually?

So I'll just start. And so, just starting probably with the team-- like, what's the demographic of the audience here right now? So how many in the construction, like, civil engineering and project management world right now here? Like, just raise your hands, please. Cool.

And what about, like, technology side, where you guys are implementing technology or building softwares? And how many are genuinely, like, have thought about implementing data and enabling better project tools, using data and historical records of your company? Has anybody explored that? Great, so a few of them.

Yeah, so I think there's a lot to learn because we've seen every industry being transferred, right? When you're walking out right now, like, your Fitbit is counting how many steps you've taken and giving the analytics upon how is your health predictions, based upon that. And so as they say, data is, right now, the most valuable resource, even more than oil, right?

That's why companies like Google and Facebook and Apple has become so valuable, bigger than all the oil and gas companies, which used to be the biggest companies. If you look at 10 years ago, the oil and gas companies used to be the biggest companies. Now it has been taken over by Google and Apple, not just because of the product they have, but because of the valuable data they have that can change industries and change economies around the world, which we've seen in the last couple of years. Like, Facebook and Google can actually do a lot than just connect people, you know?

And we've seen, even using your Google Maps, you know, it's collecting every data about where are you going and everything. So the data is the most valuable thing right now, for everybody. But what are the parameters of data? So, what does big data mean?

So big data basically has four big fours. And those are volume, variety, velocity, and veracity. So these are the four basic requirements for big data to get good insights from your data, actually. So I'll walk you through, like, what those big four Vs are for construction industry, you know?

So as you know, volume is the size of the data. And variety means how many-- because you cannot just analyze, like, five people's data, because it won't give you a really good insight upon that. We've seen some of the stories where-- I was watching this thing, and they will give you, based upon, like, 10 people's metrics, they will say that drinking wine will avoid cancers, because there was some correlation between not having cancers and, like, drinking wine, right? So it's like basically, it was very small data. And if you analyze a small data, it doesn't give you proper insights.

The velocity is important because the velocity of the data that comes through has to be, like, high frequency. Velocity has to be there. And veracity means the accuracy of the data, because you have to have a clean data. And only then, you can apply it, right, because if you don't have a clean data, it is all messed up. Then it doesn't work.

So how does that work in construction industry? So basically, construction industry, now the question is, can we apply big data to construction, actually? Like, is it even possible to do that? Well, it's possible, because we have pretty good metrics on those four Vs.

So it's a massive industry. We all know that, right? Even the presence right now here, today, in the [INAUDIBLE] more than 10,000 people from all around the world, you know? It's basically shaping our lives. It's one of the oldest trade. It's 4% of the GDP. And more than 9.5 million people are actually employed in the industry, 8.5 being, like, in the field, working directly, actually.

So that's a pretty high volume that we get from just the industry. So that's one yes to the can we do big data? Yes. We can apply big data to construction industry, yes.

So what's the velocity, right? So how often are we building things? We're, like, constantly, every second, we are laying foundations, laying brick walls, laying lightbulbs. So there is a high frequency of activities that are going on a job site. Your daily report itself, you're counting man hours of how many people were there in the job site. It's like a lot of, like, every image that you take now with the 360 cameras, you can take so much images that there's a high amount of velocity.

Now point cloud, using drones, sensors, and even the RFIs itself-- you know, we did the metrics of how many RFIs are created. You basically have 9.9 RFIs per million dollars spent on the industry actually right now. So just analyzing the RFIs itself can give you a lot of insight upon how can you-- because the RFIs will lead to change [INAUDIBLE] you know? That's how it starts impacting you. And if the RFIs are not replied, then your schedule is impacted ultimately, right? So everything is correlated here. So that's also tick there, right?

So then we come to variety. So we know that it's a highly, like-- I don't think any other industry has as much variety as construction industry. Like, every site is different, every project is different. There is so much dynamism in the project that, like, you cannot build the same building in the same location in the same year in the same way, right? There's so much variety, which is a bad thing and a good thing both, because variety brings a lot of chaos.

Now, can we find some rhythm and harmony in that chaos, because that's where the beauty is, you know? We still haven't been able to figure out, even if you're building the same building in New Jersey with the same guys, it will be different, right, just because it's such a dynamic way of working. So we're blessed to have variety, or also, we're crushed to have variety actually, you know? So that's one of the main reasons why we haven't been able to apply big data in construction industry as compared to other industries, actually, manufacturing, other industries, you know? So we're blessed in [INAUDIBLE]

However, where we lack is the veracity, actually. The veracity lacks because of the fact that we never had a really good standardized system to get the accuracy of the data. So the schedules are never updated. RFIs are never, like, prepared in a way that, actually, it can be analyzed properly. And everything is such a paper-based industry. Like, we were still dependent so much on the paper, you know, until last couple of years ago.

So everything that we have is from 1980s. That's when the Microsoft Project and P6 and AutoCAD was like-- in 1981 was when all the three things happened at the same time, you know? So whatever we have is actually, like, the last 30 years of data. But that data is actually not real data because it was all just a file, actually. The Microsoft Project file, or the schedule, was never actually updated to, like, accuracy, it was only updated to make the payments or to go to the owner and please the owner, actually. So there is actually no real, like, data that you can analyze right now.

So that's why the whole process is now taking shape, too. That's why [INAUDIBLE] is such an important thing, and all the standardization that is happening, is such an important thing, because we need to standardize to analyze it first, you know? So let's look at how and what's the ultimate goal from here.

So the ultimate goal is basically to collect the data, right? You collect the data and you clean the data up, because there is so much of a lot of data around that is messy, like your schedules are just made fake to please your owner. Because if you need to get real insights, you need to have real data actually, you know, like good one, all right?

So basically, you clean it up. And then you identify patterns. And then you make predictions and recommendations based upon the historical patterns, right? So based upon your-- which every industry does, actually. Uber will tell you what is the fastest route to go from A to B based upon the whole dynamism of how is the conditions on the street as of now, which is dynamic, as well as your historical, like, road map and all the things, right?

So this is where we should go to, especially giving recommendations. In our industry, it's also even more complex, because just giving the recommendation doesn't work, actually, because if you just say that, OK, contractor B has a high rate of failure in a certain kind of a project in New Jersey, then what is the actual-- just giving that insight doesn't work, actually. Now you have to give a proper recommendation as in, like, how do I solve this problem, actually, you know? And that's where the complexity comes in.

So getting those insights from like, do I break the contract down into different contracts? Do I decouple some of my scheduling? So those are the insights that we can get. And we're working towards that actually, right now.

So in construction, so basically I'm just-- this is some of the data that we have in construction, right? There's much more than this. I'm just showing you the slide, like these are some of the data that you know. So the information, you've got weather, your contractor's historical performance-- has the contractor done some kind of a similar work before or not, right-- and then what is the criticality matrix of that particular activity that was just [INAUDIBLE] actually, you know? So can we can we find those patterns in here to make those connections, actually?

And now, these are all independently lying in somebody's Excel sheet or somebody's file system or phones or e-mails, right? Now the goal is to actually bring them all together. And you can see that, underlying here, are all these connection and connected dots, actually, that we need to connect together to find a [? correlationship ?] between each of these, and then come up with a better insight as to how to run a project better. How can you reduce the number of RFIs? How can you reduce change orders?

So getting those insights from your historical projects is going to be the ultimate goal actually, here, right? So what's the goal? Like even if you do this now, what's the goal, then, right?

So the goal is to, then, basically apply predictive analytics to do your automated scheduling. Because ultimately, if you're building the same kind of a building in-- if you're building a nuclear plant in Japan, and if you've been building nuclear plants for the last 30 years, you should actually know exactly how we're going to build that, right? I'm sure nuclear plant people have figured that out. We haven't figured that out in our residential or commercial infrastructure, which we now need to do, you know? And do estimations as well, right?

With the click of a button, you should be able to upload a model and get your schedule and estimation to 90 or 80% accuracy, just like how Google will give you your direction and your ETA based upon your A to B direction. So you're basically able to predict durations of each activities, your cost of each activities, and then dependencies and risk of your activities based upon your historical data.

And what is the actionable insight you get? Is that you can dynamically-- but construction industry is so dynamic that, even if you predict that this is going to take six days duration, it will change, right? So everything, that has to be dynamic. And that dynamism needs to be captured in real time as well.

That's what Google does, you know? Your ETA will dynamically change based upon the traffic and based upon how are you driving the car, right? So same like that in construction industry as well. You have to dynamically be adapting to the changes that are happening in the field in real time.

So unfortunately, we're still using an antiquated system to manage our projects right now. Although we have transferred 30 years, like we now have better scheduling system, we have better cost estimation systems, you know, until a couple of years ago, it was just in a spreadsheet, you know? Like, it was just one person's desktop, lying, and it was never used for anything.

Companies will do the same mistake. We did the same mistake last year in the same kind of a building, and we will do the same mistakes again, again, again, actually, you know? That's like no-- like, how can we change that, right? So at least, like, if the same company did the same mistake last year, at least let's not do that right now, you know? So that's the whole goal of it.

And then we're always looking behind, right? Everything is like, we update our schedules just like once in a month or once in two months. And then one person is updating the thousands of activities going on in a project. And just, like, one or two person, and that's his job, you know? So we're always taking reactive decisions. But can we learn from somewhat?

And also, like for the people on the job site, he's always looking at like, he doesn't care about the schedule, right? For him, the schedule is basically something that he's going to stick on the wall and only report once when somebody actually asked him to do that, you know? He's basically just going off his own schedule, right? And if you ask anyone in the job site, they will say that the schedule doesn't work, right? But why are we even preparing the schedule and spending so much money and doing that, actually, right now, yeah?

So can we learn from other industries? Yes, we can. And we all know that. I'm that, like in many, many talks, that you've already learned that manufacturing is where we can learn from, because it's one of the closest to what we do. And so we can definitely learn from manufacturing. So what can we learn from manufacturing is that manufacturing has basically adopted these principles. That's based upon our study, actually, you know?

So basically, it digitizes the process, right, because digitization is not just for the sake of digitization. Like, you just don't get carry a PDF of your schedule, but actually digital system, where it's like BIM, right, not CAD but BIM, because BIM is more actually a real digital system, actually. And then you need to standardize it. We've been having many talks in our booth down there, actually.

And so we've seen that, actually like, in other countries-- like I'm sure there's people here from other different countries today. We've seen that, like, Germany and UK actually has much standardized BIM. And Max, you know, if from, actually, Germany.

So other countries have, actually, better standardized BIM. Like here, we are just using BIM for like, OK, just go BIM, BIM, boom boom, you know? Whereas in UK and in Germany, there is actually a really-- they do the real BIM implementation plan. They standardized the whole process of how do you name each thing, so that if you're updating it, it's across the board, you know? And you can use the same elements, not just for one project, but all across your company, actually.

So standardizing BIM's really, really important, because if you're not standardizing, then you cannot optimize it, ultimately. Because the ultimate goal is to optimize it, right, to optimize your process so you can drive value and you can get profits. You can gain more profits and you can reduce the time. And you can, like you know, then predict-- just like any other industry, you should be able to predict. Like if we are going from A to B, just like how Google will give you the four routes in your map, we should be able to give you those four routes in your schedule as well, right, because we have enough data to give you those things, you know? So ultimately, optimize your process.

And then automating it, so how much can be automated? As much as possible. Auto-scheduling, auto-modeling auto scan to BIM, or auto-updates of your schedule-- so all those automation can only happen once you basically have the digitized and standardized, and then the whole process of optimization, actually, right? So that's the process we have to follow for, like, learning.

I think we're all digitizing. That's why we're all here, right? So that's a really good step we've taken in the last 10 years, and BIM 360, like, standardizing things. But we should standardize the processes. Like, [INAUDIBLE] right now is standardizing different variables, like how you code RFIs, how to repeat schedules, you know? And I think, like, if you're going digital in your company, I would highly recommend that you guys should make a standardized plan for naming convention, updating convention, so that you're not just digitizing randomly, but there's a standardized way. And then you could all feed it into, like, Microsoft Azure, or any kind of AWS, machine learning platform. And if you just have a standardized system, you can dump that data into a machine learning thing, and it will show you patterns about what was happening, actually.

So that's the first part of my, like, what is big data? And that's my first part. So I'm going to pass it over to Max, who'll talk about what he learned in the university case studies he did.

MAXIMILIAN SCHUTZ: All right, let's switch gears. Good morning, everybody. For guys who just arrived, I'm Max. I'm a researcher at Stanford University. I'd like to give you some insights into research that we've been doing in this area as well. This is work by Martin Fisher and myself. I think many of you know Martin as well.

Right now, the title carries Dynamically adapting look-ahead windows based on project performance feedback loops. And I would like to provide you with the why we should do this. Whereas Niran provided us with the how and when and where, I would like to give you some insights why we really have to do this.

So we started out with three really simple and basic questions that we wanted to investigate. The first one was which planning cycles exist on today's most advanced construction projects? So we went to many contractors, lead consultants, designers around the world, and told them, give us your best projects, and we would like to find out which planning cycles there are on these sites.

So the second question was how reliable are these planning cycles? How reliable are your planning cycles that you have established on your sites? And the third and most important question, which ties back to the feedback loops-- are results from one cycle used to learn from and to adjust the next cycle? So, are we learning from our mistakes?

To answer the first question, which cycles exist, we did case studies on 14 projects, got a really good mix there from residential, commercial, educational, industrial projects. Went to seven countries for these 14 projects-- Brazil, England, Germany, Kuwait, Oman, Russia, and USA. And this gave us a pretty good oversight of what was going on in the world in these projects. So we really wanted to make a broad study and see how sites and planning cycles were behaving in different parts of the world.

To make a long story short, we observed four core planning cycles on these projects. Shouldn't come as much of a shock to you, the first planning cycle was project/phase planning, so project planned in a phase and project cycle. So they always was about construction strategy, means and methods, what's important there. The second planning cycle was a monthly meeting, mostly monthly, which looked at, for two or three months, always about supply chain management. We have to sign documents. Do we have the resources, material, logistics?

The third one was a weekly meeting, usually looked ahead for one or two weeks, always about production planning, the final sequence, and a collaborative commitment, in that area. And finally, the fourth one, interestingly, only six of the 14 projects planned on a daily level-- always about the final check-in, the status of completion, and the final decision of where to move today. And also interestingly, none of these six projects applied BIM methods. All of them were in the lean area. So some projects use lean management, lean construction methods, some of them use BIM. Some of them use BIM and lean together, but never in one cycle. But none of the BIM projects use the daily planning. Only the lean projects, in that regard.

Here's an example of what that looked like. On the left side, you have a two month look-ahead meeting with the project [INAUDIBLE] and project management. Right side, you have-- on the top, weekly role play meeting with foreman. And in the bottom right corner, you have a daily huddle among foreman of all trades. So they meet every morning at 5:00 A.M. in the trailer on site and go through the day. 10 minutes. Really quick. What's going on today. There is a crane over there. Be sure not to go in there. We have a delivery going in there. And then they go out and execute their tasks.

So this brings us to the next question we tried to answer. And we wanted to know how reliable are these planning cycles. So if we looked ahead for one week, we found that about 47% were on time, of these commitments in the weekly planning. 13% was too early, and 40% was too late. 40% too late is already-- oh, but we know it's construction. We have variability, so it makes sense. If we look ahead forward two months, that green area melts down to 5% accuracy. So 39% were too early, which is of course, better than too late. But it's also not perfect. Because you might not have the material on site. You might need a crane that you have ordered for in three weeks, but now you need it for this test. So it still cannot be really executed.

So 5% accuracy is you playing Russian roulette with your productivity with a gun in which 19 out of 20 chambers are loaded. And this is not a good idea to really lead your site. And if you look at for three months, we found that it's 50-50. So you might as well go down and gamble, but not do planning. Because it's really hard. And to give you more context to this-- I'm sorry. It cut off in the bottom-- that's the weekly planning cycle. So you have the development of 1,000 randomized sample of 1,000 BIM objects that we observed over the course of half a year.

And you will see on the y axis here that that's a deviation from the plan. On the lower bottom, the blue ones are too early. Green is on time. Right is too late. That's the actual execution of a BIM object in correlation to when it was planned. If you look over to the two month look-ahead, we had a randomized sample of 6,000 BIM elements that were planned in the two to three month look-ahead window. And that gives you an idea of how it developed there. So we're not talking about a few days late. We're talking about 100 days. 50 days. The mean delay was 31 days. And a month late for two month look-ahead is well--

So that's the three month look-ahead window, and these are not BIM objects anymore now. So we analyzed, again, a randomized sample of work packages. So these are several objects combined in one package. Because that's what you plan ahead for three to four months. And there you can see these are calendar days deviation here. 200 to 300 days late and too early. So this is just really a mess.

So then we went ahead and said, what are reasons for incompletion? So we know now that everything leads on a quantitative basis. Everything points towards the monthly planning where we really have to improve. And then we wanted to know why. Why are these planning cycles so unreliable?

So we did studies on the root cause analysis of each of these projects. And we've got them up at the top. So we wanted to know the top three from several of these projects from Germany, UK, Kuwait, Oman, and so on. So if we really go through these, you will see it's actually always pretty repetitive. It's always material, procurement, logistics, design and complete, design material, logistics. So 14 out of 18 root cause categories of the top three of each project, again, relate to the three month look-ahead cycle. Because these are all things that you should plan three months ahead, or even longer, of course. Because you won't touch procurement or material delivery if you're one week out. So this is something you really have to plan months ahead.

Give you another insight just specifically of one project where they use BIM for planning. They could perfectly express what's wrong. But it still went wrong. On the left side, you have what was delivered of the structure, which was currently on site. On the right side, all the parts are still missing. The story behind that is, the owner calls up the GC and says, so my supplier tells me that you have 300 tons of steel inside. Why can I not erect anything? It's these 300 tons on the left side. Who is going to erect anything with that? So you have to really align your schedule to the delivery, as well, and talk to each other. And not use a little close BIM, but also open big BIM and share collaborate with each other. Otherwise BIM won't have any use for that.

And to answer the third question, our results from one cycle used to learn from adjust the next cycle. We tried to use the Deming Cycle, known in the lean world to apply for construction planning and practice. So we tried to answer-- for those who are not familiar with it, it's plan to study [INAUDIBLE] core elements. So we said, plan to study would be our planning cycle, where study leans towards feedback loops. But if you don't act on it, it's not a feedback loop. If you don't close it-- if you study the results and don't act on it, you won't have a loop. You will stay in these planning cycles. But you won't learn from this.

So we said, OK. We do plan ahead in different cycles. As we have seen, this four cycles. We apparently go out and do something different. But we do. And we also study our results. We've seen many metrics on the site plan percent complete. Different metrics, progress reports, et cetera. But we don't act on these results. Because next cycle is going to be planned the same way, with the same methods, with the same look-ahead window, as if nothing happened. So we do not have a PDSA cycle. It's more like a PDSC cycle. Cry, complain, crash. But it's not acting. So we really have to improve on this A to get this A in this equation.

Also interesting, planning cycles were not really well connected. So each project had one person in charge to manually transfer information from one weekly to another monthly plan, keep that updated. This is a tedious unnecessary process. It's prone to error. It's automatable. And I bet none of these guys really got a degree to shift around these numbers. It's not really nice work. So this is really automatable.

Also, as I said, we observed that BIM was not used collaboratively. So they have really great BIM practiced on their sites, but also due to the complexity of the software itself, as well, they had BIM engineers on site that translated what was discussed in the meetings into the software afterwards. So none of the people in charge really used the method. And this is also not the way to go with, because then it's not collaborative and not supporting the planning process of all stakeholders, of course. If you need BIM engineers to transfer everything, there is a better way for that.

On the other hand, if we are going to have the projects, many of these lean practices are really not digitized. So you end up with information that was incomplete, manipulated, corrupted. Problems to transfer, analyze, and communicate. On one project, we were one of the other visitor groups. And the first visitor group was first. And they took a look at this planning board-- Kanban board and everything. And they took out the cards, passed them around. It was a real life planning board that people used to plan. And they passed them around, put them in the wrong [INAUDIBLE] area. They kept them, put in pockets. Laid on the table. They corrupted the whole thing within five minutes. So how in the world are they going to establish a PDSA or a continuous improvement process with that? This can also not be the final solution.

So our key observations were that the month look-ahead plans were highly unreliable. And also that there are no formalized mechanisms for feedback loops in practice. So what we have to do is just create a picture. It's that really easy. Create the feedback loops and improve the monthly planning. And then we can go on, of course. Really simple, as you can imagine. This is a lot of work. And this will take a lot of vision and change in our practices. If you're more interested, we created a video of our vision of what a feedback loop planning system should look like. We also have started a development on a prototype for that for our research project, together with Forge.

If you want to check it out, just follow the link down there, or in the recording, you will see it. And if you'd like to get in touch, I'm around. Just feel free to ask me and answer I'll answer any questions after the session. So these were some quick insights into our research and I'll pass it over.

All right. So I didn't do research on that specifically, but I worked before implementing BIM 360 Field at a GC in Germany. And we also tried to answer that question right away, what's in it for me? Because otherwise they won't listen you have to get the attention of the group of course otherwise they won't do it. And we sat down before showing them anything. We asked them, what are your problems right now? If you had one wish, or several wishes, what would you change? And then, they opened up. And their eyes opened up, and they said, OK, someone's listening to me.

And we made a whole list of the problems and issues. And then we tried to match them with 360 Field can do, with what the technology can do. And we came back, and said, OK, we looked at your questions and problems. And I think we can help you there. We can definitely help you there. This might be a thing. And then, it became their development. And we addressed their issues with this. And then it became their baby. And they really liked it. And then, it spread out to different sites, because foreman talk among each other, as well.

And I said, well I always have these perfect overviews of issue management. And I can go home at 5:00 or 6:00 instead of 7:00, because I have this issue. I can go home to my family earlier, because I have a great overview. I don't have to do reports. Reports are automated, and all that. So you always have to listen to them first and adapt the system to their needs. And then you will have them in the boat. This is a very important thing.

MAXIMILIAN SCHUTZ: Yes. Same [INAUDIBLE] I'll So we did some case studies. So the case studies basically shows that-- these are some of the products we tested our system with. [INAUDIBLE]

NIRAN SHRESTHA: And so basically, what it came down to was so there are three levels of benefit to the users. Let's go back to the guy in the field. Why is he going to do that, actually, right? Because that's probably the question right now. So I think that there are three benefits, actually. So basically, he can get his job done faster, because he knows exactly what to do in the morning. The guy in the field is always looking at, OK, can I do this? Can I do this? What should I do actually? And I think, having that information as to, OK, this is what I'm going to do today, actually. So he gets his job done faster.

And down to the field, where now, everybody has to do their daily reports, right. So when you tie the daily reporting system to your planning system and your CPM scheduling. And that becomes, I have to do my daily report, but when I'm doing my daily report, actually I'm also updating my schedule, actually. I'm [INAUDIBLE] what I did actually.

So and why you do a daily report, because you're obligated to do it. You're getting paid based upon your daily reporting, actually. It's also a compliance thing. So it basically gives you the right information at the right time. And then, gives you the right time, so you know you're saving time. You can do your daily report by a couple of clicks of a button rather than going to your office and then starting to do that, actually.

And then, the other benefit is that you're ultimately being accountable for what you do you. And as you get more involved in the planning session, you're not being imposed-- the scheduler is not imposing a schedule on you, actually. You are actually putting your voice out there, because in a [INAUDIBLE] planning session, you have all your subs in there. And they are saying the durations. So they can take that accountability of, I said I'm going to do this in six days. So I'll have to do it.

So that's why I think the field guys are-- we've done the case studies on why the field guys have been very happy to adapt the system that we built. And then, for schedulers like, why are they going to use it? Schedulers or estimators, basically, or VDC managers, because they're able to create schedules faster. And because of these historical schedules, they're able to now also get schedule updates faster.

So we've done case studies where we've decreased the number of schedulers, because the field guys are more involved in updating the schedules on a more real time basis, actually. And ultimately, for the high level people like the decision makers, you're always looking forward rather than backward. So I think there's three levels of value across all of them. And I know that the value definitely has to come from the bottom, because ultimately, if the field doesn't adopt it, you're just going to drive the solution x-ray. So, yeah. I think we've seen that growing and just digging further and further is going to be the goal. Yeah.

So I'm just going to quickly show you some of those-- yeah. So I was here. So consciousness learning is one of the principles of your lean management. And so conscious learning is basically, you learn from what happened. And then, you're always improving. So same like that, you can actually apply machine learning to that, actually. As you have standardized data, you cannot stop optimizing and giving the predictions of OK, can I predict my next four weeks look-ahead more accurately? Is our [INAUDIBLE] going to be more accurate based upon historical data I know? So those accuracies is what we're getting to.

So this is a dashboard that basically shows the risk metrics of each activity in the field. So and what are the root cause of those metrics? So you can see red, blue, and yellow on these activities. And what are the root causes? Is it the weather? Or is it missing information? Is the RFI missing is going to drive that activity more delayed? And transferring that information back to the guys in the field where you have a more dynamic way of planning and making your products faster.

So these are some of the metrics that we used. And as we said, basically it used to take a scheduler 60 days to collect the data, analyze, and make the earn value report. And now, your earn value report or even all your risk reports are always looking backward. Now you can look forward and make more informed decisions.

So some of the case studies we did basically-- so right now, I'm just going back to what we do. Right now, we have around more than 120 projects growing globally primarily in New York, because we're based out of New York. And in San Francisco, Canada. And some in Brazil, as well. We've got some of the people who have tried our software here, actually. Some of them have done it in China. And, yeah.

And then, so these are some of the case studies that we did with some of the projects. So the World Trade Center is one of our customers right now. They're doing the construction of the subway station underneath it right now using our software. And then the Sandy Renovation Project which was the Hurricane Sandy that happened in New York City was one of ours. So even down to the level of just renovating your buildings, it can utilize the system, actually.

So and we did some case studies. So and we were able to show that we were able to bring the project duration by 20% faster. And just in one year, we were able to save $400,000 dollars for this customer of ours. This is a mid-sized construction company based out of New Jersey. And these are all-- below are some of the other people who are using our system right now.

Yeah. And so the metrics also shows that how do we actually ultimately save is basically we're saving seven hours per day for each worker on the job site. Because he has the right information in his hand. And also, he can immediately transfer any information back to the office just by a couple of clicks. And ultimately, seven hour translates to seven hours for who? For the super, for the project managers, for the schedulers. And so, these are all of the people who are getting the benefit of using our system, actually.

And the ROI across the life of the project, as we get higher, all the subcontractors also start coming into the system. And that's the ROI for the overall project across for a typical $100 million project, actually. So now, I'm happy to take more questions. Or, me and Max. Yeah.

AUDIENCE: [INAUDIBLE]

NIRAN SHRESTHA: I'm really happy that you're still awake. It's pretty early.

MAXIMILIAN SCHUTZ: You guys have the first class.

NIRAN SHRESTHA: Exactly. Yes. It's a most unfortunate spot.

MAXIMILIAN SCHUTZ: Yeah. I will probably ask questions if you don't have questions like this. Do you have a question?

So I think I saw some of the people who were saying that they are applying data to their company. And I just want to learn, how are they planning to use data and make better practices in their company? I think there were a few people who raised their hands on that. Yeah.

AUDIENCE: Just tracking that data, it's on task. And it's all of equal importance. So there's [INAUDIBLE] This is a great presentation. [INAUDIBLE] I don't think anybody clapped yet. Are you guys done?

MAXIMILIAN SCHUTZ: Yeah. We're good.

[LAUGHTER]

[APPLAUSE]

Thanks.

AUDIENCE: So [INAUDIBLE] I'm going to download later when I'm awake more.

NIRAN SHRESTHA: Good call.

AUDIENCE: [INAUDIBLE]

NIRAN SHRESTHA: So, yeah. You're saying, a task data [INAUDIBLE]

AUDIENCE: So we'll throw [INAUDIBLE]

It's started tracking these metrics. Otherwise, [INAUDIBLE]

NIRAN SHRESTHA: Exactly. Yeah. And then, standardizing will be definitely important, as I had mentioned earlier. Digitizing and standarize and optimize. And once you standardize across not one company, but standardize across the whole industry, then you can predict everything after that, actually. So that will be a valuable thing to do.

So if you have some metrics, you can send me an email and we can discuss it, as well. Yeah.

MAXIMILIAN SCHUTZ: Great challenge for the [INAUDIBLE] is that we will have to standardize a lot in order to really learn from it to have structure data. And when we implemented BIM 360 Field, for example, the project managers used it the most when they were able to customize what they wanted, categorize the issues, and if they could, come up with their own categories. It was their own system. They wanted to express what they felt is the best term for that. And it became their customer system.