Machine Learning for Construction Safety: A Construction Project Manager’s Perspective

COOPER DARLING: So we're really excited to be here today to talk to you about something we've been working on. Autodesk invited Layton Construction to be part of a pilot program or a beta program for machine learning that we'll be discussing today. So we're excited to get started and wanted to do some introductions

So my name is Cooper Darling. I'm an assistant project manager for Layton Construction. I've been in the construction industry about six years, four years with Layton Construction. Previous to that kind of the typical internship and worked for another general contractor in a different area, but kind of been involved in some mountain projects, resort projects, office building projects, restaurants, and things of that nature. So that's a little bit about me.

AARON GETZ: Great. Thanks, Cooper. My name's Aaron Getz. I am a business consultant with Autodesk. I've been with Autodesk for about 3 and 1/2 years now. And prior to that, I worked for a general contractor for about nine years. I did both project management and was a BIM manager there as well.

I've been fortunate enough to be working with Layton Construction for the past three years or so. So we're going to talk to you guys a little bit about our journey, what we've gone through with the things that we've set up, and a lot of the stuff that we've learned along the way. And we're excited to share that with you.

So that's a little bit about us. We want to get to know the audience just a little bit here. Quick show of hands, how many are from the construction industry here? The vast majority. Any from the architecture engineering or design? OK, quite a few there. And how about ownership side? Some from the ownership side as well. Very good.

That helps us understand who we're talking to. I believe that the concepts and topics that we're going to be addressing today will be relevant and beneficial for everyone here today. So we're excited to share this with you.

COOPER DARLING: So our agenda today is we're going to go through like we talked about our backgrounds, talk about like Aaron mentioned where Layton Construction has been and where we are going and what we describe as our dilemma. A lot of data in the construction industry, a lot of information as it pertains to what this class is about, our safety and our subcontractors, talk about our dilemma, and then talk about some of our processes and what we consider a great solution to this as we teamed up with Autodesk to solve our dilemma, and then talk about how does that affect us as project managers or how it affects maybe the end result for the owner or even being involved with architects. But really how does it affect me personally, for you personally on a project level? And then we'll kind of go into what are the possibilities of machine learning and where BIM 360 IQ could potentially go in the future.

So a little bit about Layton Construction. We've been around about 65 years. We've got offices all over the country, kind of spread out a little bit. But primarily, where we have most of our work is in the Mountain West region. One of our larger sectors is health. We're one of the largest health care general contractors around. We do a lot of municipality work, office buildings, things of that nature correctional facility. So those are kind of the some of the sectors that we work in, definitely not a comprehensive list. But that's a little bit about us.

We pride ourselves in having a culture of partnership through our relationships whether it's through the owner or through architects or through subcontractors to try to have that collaborative work environment. So that presented itself well with working with Autodesk. And in safety, so we consider ourselves a safety leader, and we want to continue to push the envelope of what safety is in construction and help keep subcontractors safe and us when we perform stuff as well.

This is just a quick view of where some of our main offices are, as you can see from Hawaii to Florida. That's kind of where we're at.

OK, real quick. So I want to get some input from you guys in the audience. Who here has a smart cell phone? I'd imagine probably pretty much everybody. Who has like a smart smartwatch? Those are a little bit more.

So just anyone want to blurt it out? What's one of the things that a smartwatch helps you with?

AUDIENCE: Fitness.

COOPER DARLING: Fitness.

AUDIENCE: Sleeping.

COOPER DARLING: Yeah, sleep tracking.

AUDIENCE: [INAUDIBLE]

COOPER DARLING: I didn't know it could do that. All of those are examples of just capturing data. Obviously, I don't really use my watch too much for fitness. But it can do that. It gathers that data. It still tells me how many steps I take, all that stuff. However, if I don't use that data properly, it really doesn't do much for me. I still can see it.

But that's what we wanted to talk about here in this slide is we've got different examples in the construction industry of how we gather data, whether it's through BIM 360 Field with an iPad or if it's Austin here using the drone-- that's Austin up there taking drone footage of our projects-- or data clouds. Those are all examples of how we gather data. And there's a tremendous amount of data that is swirling through the construction industry, whether it's in the design side, ownership side, subcontractor side.

Our dilemma, like I mentioned, is just we have a tremendous amount of data. This is a representation of the data as it's shown BIM 360 Field and what typically our project teams will see, which is lists of issues or lists of checklists. And as we're talking about safety here today, we make positive or negative safety observations. All that data is there, but we wanted to have a way to extract that data and use it for our benefit.

As well as, this is a picture of one of our stretch and flexor or pre-start huddles that we do corporate wide and on every project. This is one of our Hawaii projects. But we have many subcontractors. And each of those subcontractors has a culture and has information and has different ways of working. So as we analyze and work with different subcontractors, we continue to gather data on how they work. And so that's another part of the dilemma is we just have so many different companies that we're working with.

So on an enterprise level or a company-wide level, we began determining in what originated in our safety department, corporate safety department, is there was a need from our CEO, his own message out to the company is we need to improve our safety culture. What was created from that is a committee, which I have the opportunity of sitting on as well as the national safety leadership team. This team is comprised of people from all over the nation within Layton Construction. I think there's about 12 to 16 members, and they represent different parts of the company.

But what the group does is the group comes together and determines what are important things that Layton Construction wants to be tracking as far as our safety program. They come together and determine what processes and programs or procedures are important. So that's where we determined that to help us gather data that is useful, we needed to have a standardization for our company internally and our project teams on how to accumulate that data, so that the data could be used more effectively.

AARON GETZ: In addition to the safety leadership team that Layton created, we also created a quality committee. And this was a similar type initiative as the safety. And that was that we wanted to establish a standard. So we want to improve safety. We want to improve quality.

Well, what does that mean? How do we do that? How do we accomplish those things?

So that could mean, hey, this is a required meeting. Or we have a certain procedure or a checklist or process that we want you to follow. So we created, again, at the enterprise level, a group with which we can talk about quality. What's happening on our job sites? What are we seeing today? How can we win more work? How can we improve our relationships with our owners and our design teams?

And so this was really a foundational piece. When we started three years ago, it was creating these committees and building programs and processes, establishing the standard. That has to happen before we look at data. Without those users, without that foundational base, we don't have data to look at.

And as you see here, Layton experienced significant growth. We let the monthly active users and all the BIM 360 activity just follow. We were more concerned about our processes and our procedures and rolling out the quality program and the safety program. And we found that this really just followed.

But again, this is really, really important. We're going to talk about machine learning. We're going to be talking about data and what it's telling us, the story behind it. But there's not much of a story if we don't have people putting in that data. So having a foundation of high user activity is very, very important and something that every organization really needs to experience before they can reap the benefits of analyzing their data.

So those other solutions that we address, the safety leadership team and the quality committee, those were enterprise solutions. Those were big, high-level solutions to address these challenges. And really what we want to focus on today is how does this help the project?

Cooper is a project manager. He's asking himself, how does this help me? That's great we have a quality committee. That's great we have a safety leadership team. But how does it help me on my job?

Because, today, I'm worried about the stuff that's right in front of my face. I'm most concerned with the fires I need to put out right now. And I don't have time. I don't have time to comb through all this data and to access all these different platforms and databases. There isn't time to do that. And more importantly, I have to manage subcontractors. I have a team of people that I have to point in the right direction, so that we can orchestrate this project together.

So now, we want to talk about at the project level, how we can improve at the project level.

How many people have heard of BIM 360 IQ by show of hands? Great. So we have some people that have learned about it.

Cooper mentioned previously that Layton Construction has had the opportunity to be a beta tester, part of a pilot project. And we've been experimenting with this product. We've worked closely with the product team and have enjoyed a great relationship with them.

But this, we feel, is going to fill that void at the project level and how we can improve, how we can get that data and that insight at the project level. It's going to address the immediate needs of the project team on a daily basis. And I think that's what's most important. It has many other benefits as well that we'll elaborate on.

But it's going to prioritize our high risk items, our subcontractors that are in need. It's going to comb through that data. It tabulates the data. And more importantly, it provides a predictive analysis of that data. And that's huge. That's a first. And so we're real excited about that.

And here's what we have found while using BIM 360 IQ. We don't have to mess with extracting data. That's not a challenge anymore. We don't have to worry about the format.

And we also don't have to worry about latency because it's live data. It's a web-based application. The data is always current. It's up to date.

As I mentioned previously, it helps the project team focus on their priorities. What's most important today?

And then one of the most important things is it helps us save time. And that is a key element. Most project teams are scrambling. They have a lot to do. Most contractors are running as lean as they can. They want to provide the most benefit for their owner. So they have to run lean. They're left with a smaller team so we have to look for measures and ways with which we can save time.

So I want to quickly show also just a little comparison here. This is a typical BIM 360 Field report. Who's seeing something like this before? So probably a lot of us.

We set up a report. It exports for us. And we see the companies. We see the status. We see the author, the due date. We can understand what's happening with that specific issue.

Now, this is great data. This is really important and relevant data. But what happens when we start to get into the hundreds or thousands of issues on our projects? How do we comb through this? How do we prioritize this? How do we know what's most important? Which ones do we need to work on first? How many does XYZ subcontractor have in here? Those are all questions that we have. And as I mentioned, we don't have time to go through all this data.

So with BIM 360 IQ-- you can see a dashboard here-- it helps us prioritize all of that. We don't have to perform that manual tabulation any longer. And we have a quick glimpse at what's happening on the job site level.

It helps our project teams. But furthermore, the executives, when they come on site, they don't have time to review all of that data either. We need to give them a quick glimpse into what's happening on our project sites. And IQ is providing that quick glimpse, that health check, to understand what is happening today.

And as I mentioned, the predictive part is just huge, massively beneficial, because it helps us see into the future. It helps us identify risks or potential challenges that we might have based on historical trends and what's being input into the system today.

COOPER DARLING: So recapping a little bit. We've seen kind of what the older version is or the previous version and what the future can be. We want to talk a little bit about how that can affect a project manager's or project team's day-to-day.

So up on the board, I have listed two different categories; one for safety, one for quality. That's definitely not all the things that a general contractor has to worry about. But that's what we're going to discuss today.

And then off to the side, you can kind of see a graph that comes from BIM 360 IQ. And just a quick explanation of the graph is on the left-hand side of the graph, you have the fatal four, which is what the industry has deemed the fatal four, the most common injuries or incidents that will cause death. And then you have a list, and this slide here doesn't show the whole thing, but it a list of different categories of different types of injuries or different types of observations.

So as Aaron mentioned, it will give you a predictive output and let a project team know where your risks are. And listed under safety, we have our daily huddle, which as I showed a slide earlier of our subcontractor daily huddle or pre-start huddle where we do our stretch and flex and we do a mini coordination meeting.

So where you can capitalize on that data and use IQ to change the culture in your project is you determine what issues are there as far as safety. And you talk about those in the huddle. You share them with the subcontractors, so that their minds are on that or you correct issues at that point. And you know which ones are the ones you need to focus on because they're the high priority.

That can also help when subcontractors are doing their pretest plans. We have a policy with any subcontractor that we work with they need to do a pretest plan. So we also audit those from time to time on a random basis from a project team level. And we can help them focus on those items that we hope they're discussing in their pretest plan so that their whole team is aware that that's not a high risk for their company due to their previous trending.

Or safety focus meetings that we do on our projects. We do a weekly safety focus walk. And we make observations that actually help drive this data and help continue with the IQ and the machine learning and the predictive part of it. But it also gives us a place where we know to go out in the field and start looking and start correcting those items.

Or staff meetings. Making sure that proper action is put into a project team member's to do list that they either follow up with that company or whatever it is.

Same idea with quality. There is a quality component to this. But again, the example here is that it can be used for various different facets in our industry. So subcontractor meetings, it can be discussed in subcontractor meetings. Hey, you know, we've been seeing that we're missing all sorts of things on our rebar layout. You need to be focusing on that on this next deck pour that we're doing, whatever you have.

Project reviews, like Aaron mentioned with executives, so they can come out and understand where we're struggling at a glance. And they can say, OK, well, we want to talk about these items in our project review.

Again staff meetings comes up. Or inspections, quality control inspections, we use BIM 360 Field heavily for that to conduct our quality control inspection. So if we continue to see an issue and IQ helps us understand where that issue is it can be something that's added to our quality control checklist and say, well, we've continued to miss this over and over again. We need to make sure that this is in our quality control checklist, and we mitigate that, or a safety checklist for that matter or anything like that.

AARON GETZ: So real quick before we move on, the beauty of all these different processes that we've outlined here is IQ really gives us an agenda, a tailored agenda or message to our crews, to our teams, of what's important, what's happening on our job site. So just like several of you have smartwatches and if it told you, hey, you need to run further today. Well, that's great. I don't run. We need to know what we're doing. We want the information, the data, to be tailored to us and our needs.

And that's what IQ can do for us. It tailors to our needs. So rather than running down a list of topics to go over in your pre-start huddle with all your subcontractors, let's talk about the real risks on this job site today, what are we experiencing. So it's an automatic agenda for us. And we can share that. And I think that message will be much better received by our project teams because it's a specific message. And it's needed for them. Please.

AUDIENCE: So what exactly [INAUDIBLE] categorize issues. [INAUDIBLE]

AARON GETZ: So we're about to do a live demo. But, additionally, we are fortunate enough to have some of the product team in the back of the room. I can't speak on behalf of the product-- I'm supposed to repeat the question. The question was what predictive analysis is being performed by IQ. And I invite Manu if you'd like to come in and address that.

MANU: Sure, [INAUDIBLE]

PAT: [INAUDIBLE] identifying which subcontractor [INAUDIBLE] and there's another algorithm that looks at that [INAUDIBLE] and at what point do they [INAUDIBLE] I think we've got five or six machine learning algorithms that are nested within [INAUDIBLE] identify themselves [INAUDIBLE]

AARON GETZ: So thank you very much for that explanation. I hope that helps. We're going to dive into the product and do a brief demonstration as well. And I think that will help answer some questions. Are there any other questions moving forward? All right. So forgive me while I get into the product here.

So when we first log into the product, we see-- if you have--

It's a really cool dashboard, I promise.

COOPER DARLING: Maybe it's on another screen.

AARON GETZ: I don't know why it turned off mid flight there. So when we log in, if we have the correct permissions and the correct role, if we're an account admin, then we can see what's happening within the entire organization. So this is a quick glimpse into Layton Construction. We have 364 active projects.

Down below here we have some projects that it's identified that needs some support. Below that we have a leader board. And that leader board is telling us the opposite. These projects are good. It's not identifying risks at the moment.

And then we have a map over here. As you can see, it's just showing us geographically where all the projects are located and what their status is. Green meaning low risk. Yellow, medium. And red being high.

Now, we can drill down deeper. So if I click on one of these projects, it's going to take me to a specific project within the account. So I'm looking at this particular project. And now, instead of different projects here, I'm looking at companies. These are subcontractors on this particular job site.

And it's analyzing that data. It's reading all of our issues. It's understanding what's happening with our inspections and our checklists. And it's identifying that some of these subcontractors need some attention. They need some support. It's going to give us some very easy to read graphics as well.

We talked about issues. And when you have thousands of issues, what do you do with that list of thousands of issues? Well, if we scroll down here, I can see exactly on this particular project what some of these high risk issues are.

OK, I'm not going to pronounce all of this correctly. Underside of this particular beam is not painted like the rest of them, blah, blah, blah. Now, what's neat about this is when I change its priority, it's going to learn from that. So for example, I can click on this particular issue here, and if necessary I can change that categorization from a high risk to a medium or low risk. I have the ability to do that here. So it's interactive as well. And we can participate in this process and help those algorithms become even more clever and more smart.

Now, we wanted to showcase the safety portion. So for example, right up here, if I click on Safety Risk, this is the dashboard that Cooper was showing in that previous slide. Now within this dashboard, we see some really neat things.

It's saying this project has high risk today. Right next to it, these are the high-risk subcontractors. We also have positive observations. We know that that's equally important. There are subcontractors and people that go out of their way or above and beyond to perform and to exceed our expectations. We want to know who's doing that. This is so important. We want to know which subcontractor we're going to partner with for the next project. This is going to help us.

We talked about housekeeping items. Housekeeping is such a huge contributor to other injuries and issues and incidents that happen on job sites.

And then we have a trend where we can track the fatal four that Cooper mentioned. A high percentage of all injuries, construction related injuries, are caused by these fatal four. So what's happening on this job site? Are we improving? What can we improve upon?

And to understand what, we can just scroll down here, and we can see our subcontractor safety risk. This is a real easy to read heat map, identifying which of the fatal four we're struggling with, which subcontractor needs to improve. And then as Cooper noted, on the right side of the screen, we have in yellow all of our different hazards that are associated with this project and who is struggling there.

So once again, this saves time, because as a project manager or project team logs in, they can quickly see what's happening. It's fast. It's direct. It's their data. And it really helps analyze what needs to be talked about in our daily huddle today.

How about our subcontractor meeting? My executive, my director's coming on my job site today. He wants to know how we're doing. Have we are improved with our fall risk? Or whatever it may be. This can provide that insight.

Any questions on the product itself? OK, let's hop back and hopefully it doesn't kick off again.

So we're really excited about IQ. We're excited about where it's going. We're excited about the insight and the help and assistance that it's provided to the different job sites already. Very, very exciting.

Again, we can't speak on behalf of the product. But these are some of our hopes right here that other platforms will be included in this analysis and that we can improve our insight and our information gathering by looking into other platforms, not just BIM 360 Field, but let's move beyond that. And let's move even beyond Autodesk products into some of these other partner data platforms.

And then let's put it in a single platform. And let's let that machine learning, let's let those algorithms work. And then let's view that data. And again, viewing that data, we might see in the future that data is available in other platforms and whatever your organization's BI tool happens to be. They have Domo, Tableau, PowerBI listed here.

But there is a lot of possibilities. And we're really excited about where this product can go. It's given us a lot of insight, a lot of assistance. We're still at the pilot phase within Layton Construction rolling this out. And so we're excited to roll this out to a greater audience and reap those benefits.

COOPER DARLING: I don't really have anything else on that other than it's really fortunate that as we probably heard around AU this year and last year a lot, the Forge platform and that others are able to create APIs or ways to tie into BIM 360, the BIM 360 suite. And it's just going to become hopefully in the future something that's very, very useful for us. And as we're still in the beta, it's already proven that for us in our project teams and company wide.

But anyway, we wanted to go on to some Q&A. Just open it up for you guys. You have anything? Go ahead.

AARON GETZ: In the back, please.

AUDIENCE: So the analytical side, is it also challenging the quality of the data. Is there any part of it that is looking at the quality of the data?

AARON GETZ: That's a great question. I can take a stab. Or we could let someone who actually developed the tool answer that much better. And we're lucky. They weren't planning on being here. And so we're very fortunate they happen to be in the session. And I'm certain they can do a much better job than me in describing that.

COOPER DARLING: So the question was the algorithms, are they analyzing the data, the quality of the data in the issues or the checklist? Right?

AUDIENCE: Yeah, it's challenging the data. For example, is all of your issues out there are people focused more PPE, for example, versus fall prevention. Is it actually looking part of it or is it taking at face value that data in there is what on the job site?

SHUBHAM: So if I understand correctly, the question is that given an issue, if it's fall risk or if it's PPE related, how well does it translate to what's happening on the job site and is it's really adding value in that way?

AUDIENCE: Well, we see a lot of PPE issue, but we're not see, for example, a lot of fall protection or trenching excavation issues. If you ever go to the job site with safety, corporate safety, we will see anything people just walking by. But catching the PPE is not--

SHUBHAM: I see, and that's a great question. So in this case, like obviously data quality matters a lot and what gets captured into the system is what gets reflected on the dashboard. So for example, if PPE related issues are being captured, the data is reflected into the system.

What we also try to do is one of the great ways of capturing really good data is using great checklists. And one of the things we are trying to do is create user checklist by democratizing checklist adoption on job sites. So part of IQ is what we saw the risk analysis page. But we also have two other dashboards that go along with it, both for quality and safety.

And one of the things we try to show there is the kind of checklist adoption that is happening, and for the checklists that are being done, the positive and the conformance rate and the compliance rates for that to help to understand what kind of data is being captured.

COOPER DARLING: So that ties back into that like we talked about the processes and procedures of making sure that our teams on those weekly focus logs are looking at what they're supposed to do, because that has been one of the issues that we started out with is-- we kind of call pencil whipping, and I think that's a pretty generic term. But you start seeing people just, OK, well I did my safety walk, and I noticed 10 people not wearing their safety glasses. So that's what we're trying to get away from.

So I think you had a question up here in the front.

AARON GETZ: Hold on one second.

PAT: I just wanted to do a quick followup. It was a great question. In addition to what Shubham said, one of the things we do look at is overall adoption. So we have a flag that we call a low adoption risk.

So if people aren't recording any safety issues, we don't assume that there are no safety issues. We simply assume they're not recording safety issues. So I think that's what you were getting.

One of the interesting opportunities-- and this is future. You probably saw that Autodesk announced a lot of integrations this week. We also invested in a company named Smartvid.io. What they're doing is pretty interesting, which is they're doing image recognition. So you can cull through all the safety images. And they're trying to automatically detect safety hazards. And pictures don't lie, right? So as long as that picture's captured, then you can see it.

So I think it's a great question. And I think getting people to use these systems and capture data is the first step for any GC. And you're way ahead the game if you do that. And in the future, it's all of our jobs to just make it easier and easier to capture data in any way possible, whether it's text, whether it's images or any other method. Great question.

AARON GETZ: And if I could just add one thing to that, that's what was really our focus at Layton Construction. And that was to build a process first. We wanted to make sure we weren't just looking for PPE issues. We wanted to identify lots of other issues as well. So we identified a process first, tried to change the culture before we looked at the data. So just to follow on and reiterate what Pat said. We had a question up here. Please.

AUDIENCE: I was just curious who is owning the data? So that if something comes up in a lot platforms. So I think with this one 360 being the topic, there is some sensitive data--

COOPER DARLING: So the question is who owns the data in BIM 360 Field. That would be Layton owns that data. That's our database. And it's our enterprise software that we're using that we pay for. So it's our information. Back there.

AUDIENCE: So [INAUDIBLE]

--has to get standardized input based on standard checklists, is that it? And my followup question is then what's the quality of the analysis based on various inputs? Let's say for example, someone [INAUDIBLE] in fall hazard versus fall of some other kind. What's the level of training?

SHUBHAM: I think this is being recorded. And they ask us to repeat the question on the mic so that it gets captured. And the question was that how sensitive are the models? Or depending on how the detail is being captured, how would the results vary?

And in this case, we are using natural language processing model-- you bring up a great example of like fall risk, fall hazard. So in this case, like we are building models that use natural language processing to classify issues. So instead of just looking for individual keywords, these models look for presence of multiple key words and look for the semantic context in the text that's being captured.

So of one of the experiments we did when we were developing these models for quality was we took a set of issues. And all those issues, they had nothing to do with water risk. And we added the word water to all of them and then ran them through our model. And our model did not tag them as water penetration risk because it was actually not there.

And our models simply look for the presence of individual keywords. Even if the word fall risk is not present, it could still tag that issue as a fall risk issue if it talks about someone being close to a leading edge or if it talks about improper use of ladders.

MANU: Another aspect of it is that it's not just the data that is within the text itself, we actually spent time with experts to help us label, in a sense, provide the information of what's in this text. What does it mean? What's the context?

And then we did this with a selective, a specialized set of issues. So we could actually extrapolate out to larger, broader set. But the idea there is data that is infused with expertise knowledge, and then the model is in itself is trained to capture the expertise knowledge. It's not just the structure of the language itself. So that captures more of the variance in a sense.

AARON GETZ: You had you had a question about the checklists or whatever processes that we put in place when we start. And I'm going to try to answer that. I'm trying to remember the question. But that really depends on the culture of the company and what you're looking for. It takes a little bit of a different spin per your safety program and what's needed within your industry and what it is you're trying to achieve. But right now, the software is looking-- IQ is looking at issues and checklists.

COOPER DARLING: Please.

AUDIENCE: [INAUDIBLE] you guys just have yourself doing the quality data entry [INAUDIBLE] we allow the contractors to do [INAUDIBLE] as well and allow them to bring those up [INAUDIBLE]. Now do you give them permission to do the same thing or [INAUDIBLE]

COOPER DARLING: So the question is if the subcontractors were able to input data or if it was just Layton Construction, correct? That's a really good question. And it's kind of a yes or no on both of them. So we require subcontractors to do a weekly safety inspection for themselves.

It is still very new for subcontractors to-- well, I shouldn't say very new. It's still relatively new and it's hard to get adoption that way. However, we did have them do their own safety checklists. And that can feed into the data. So it was both.

AARON GETZ: And for quality-- Cooper is speaking about safety-- for quality, absolutely we have subcontractors there. That's the premise of the entire quality program is to make those putting the work in place responsible for that work. So we welcome and solicit as much participation from subcontractors as possible.

COOPER DARLING: Other-- go ahead.

AUDIENCE: We would be looking at [INAUDIBLE] IQ, currently use a checklist [INAUDIBLE] or what's the front end setup [INAUDIBLE]

AARON GETZ: That's a great question. So what's the required setup to begin using IQ? For us and for Layton Construction, it was partnering with the product team to be a beta tester. And if you have questions, we will have some contact info up here afterwards with which you can use to get in contact with them.

But for setup, no, it's easy. If you have your processes in place, the data is there, there is no set up, at least on the customer end. I can't speak for the product team. But it's there. It's functioning. It works. Very simple. Any other questions? Please.

AUDIENCE: So presumably [INAUDIBLE] but then, [INAUDIBLE] an example, [INAUDIBLE] another company owns their, another company owns their data, silos of data [INAUDIBLE]

AARON GETZ: Good question. And so the question was, if everyone owns their own data then we have a siloed effect and how do we reconcile all that data? For Layton Construction specifically, all of their subcontractors, their designers, their owners, they are on the Layton enterprise system. So we don't have that silo that you're referring to.

And I think typically speaking for a BIM 360 product, that's the preferred method is that there's one account that's utilized for that project. And to date, Layton has been quite successful in managing that data and hosting that account for their projects. I hope that helps answer your question.

PAT: I think [INAUDIBLE] I probably need a microphone to hear me. So it's imperative that our customer data is private. [INAUDIBLE] They own their data. It's their data. We have not right to access it without their permission. However, these models that we're learning, once they're training on the data set, they can be applied across [INAUDIBLE]

We've worked very, very closely with plenty of the leading GCs in the United States and the UK. And all of that knowledge, initially we go through our models for sense of verification and training and [INAUDIBLE] sessions we did before we even had these products [INAUDIBLE] So we can learn [INAUDIBLE] the integrity of security and privacy of each customer. And none of our GC customers would ever want their stuff accessible to others. Does that help?

AUDIENCE: Yes.

AARON GETZ: Thank you, Pat.

COOPER DARLING: Other questions?

AUDIENCE: How long did your guys' team utilize the data? Is that something your utilize fairly often or was it like it's on the dashboard?

COOPER DARLING: The question is how often do the team utilize the data? Well, it's still ongoing. So that's the answer is it it's you know a typical rollout where it gets a little bit slower at the beginning, and we try to exponentially increase.

This picture up on the screen is actually the project that we did most of the testing on. This is where a lot of the processes and procedures were implemented. However, it was a weekly meeting that we had with the project team. Some of them didn't really even know what was going on and that was kind of strategic so that we would understand a little bit of how they're putting their data in. But it was done throughout the company wide. I mentioned the national safety leadership team. We established the process and procedure and we started looking at it on an enterprise level.

AARON GETZ: To add on that, it's really a mind shift. So we're used to showing an Excel spreadsheet to a superintendent and telling him or her, hey, these your issues. Now, they're looking at a dashboard. And so they have to gain confidence in that. They have to learn to trust that. They have to understand and navigate that process. And that's what's ongoing right now and we're working through that with Layton. Good question. Any others? Please.

AUDIENCE: So you said this project was recently completed. Did you see any reduction in safety incidents on this project versus what your would see normally with another using a classical method versus this tool?

COOPER DARLING: Yeah. The question is did we see a reduction in safety incidents on the kind of what we call the test pilot or the test project. And the answer is, yes, I believe we had one doctor case on the project and then one-- well, two incidents, which I can't really speak to the all the whole data of Layton Construction. We have some people from the safety department in here that could speak a lot better to it. But we did see that reduction. And it was talked about in the national safety team as an example of reducing those low safety incidents on the project.

AARON GETZ: Furthermore, we're looking into financial success. And this particular project where we employed all these processes, it was a very financially successful project, which is also important. Safety is more important. But we're seeing lots of synergies between quality and safety.

And when all of those things line up, everybody wins. And we were able to take home some money at the end of the day. So that's a fantastic thing. And this product was a perfect showcase for it, because we saw those things fall into place.

COOPER DARLING: But that is a great question. I mean, that's part of the challenge. And another dilemma is making sure that we have some sort of data to go back to the higher ups that you know the proof is in the pudding really when you see it. So, yeah, that is where we're at right now.

AUDIENCE: How is it [INAUDIBLE] it seem like [INAUDIBLE] but how do we go from there to, say, [INAUDIBLE]

COOPER DARLING: So now you got to walk back to Shubham, because we don't know.

AARON GETZ: I can give you my interpretation, but again, if that's OK, Shubham, I'll refer to-- I'll defer to you.

SHUBHAM: So, yeah, this one is a really good question. How are quantifying risk on the job site. So we start with looking at the issue text and by understanding like what's happening at the issue and checklist level and from there filling in the gaps in the data, as you saw filling in, predicting the fatal four and the other hazard categories associated.

And then from there, it's kind of now getting into the secret sauce behind how the risk models really work, but these risk models are very heavily influenced by the existing data that we already have that quantifies given these barometers what kind of risk have we seen in the past. So we've tried to do that and derive statistical models from that to predict the risk. And so we understand the relationship between the fatal four and the risk and the present hazards in the list and checklist adoption in the list. And all of those become factors into that equation that is used for predicting risk.

COOPER DARLING: Shubham, you should come right the whole algorithm up on the screen. That's probably proprietary as well. Maybe don't do that.