Augment It: Leveraging Data And Artificial Intelligence In Design And Construction With Mehdi Nourbakhsh, Ph.D.

COGE 185 | Artificial Intelligence

What if you can mimic your top performer’s capabilities, multiply their productivity, and not have to worry about attrition or retirement. In this episode, author, speaker, and the CEO of YegaTech, Mehdi Nourbakhsh, Ph.D., goes into detail on the different branches and features of AI and shares insights on how artificial intelligence can solve problems in design and construction, and bridge the gap between these two different worlds.

A special offer for Construction Genius listeners:

Mehdi is offering a signed copy of the book to the ten listeners who download the free book chapters or the workbook here: When they download, put “Construction Genius” in response to “Where did you hear from us?”

Watch the episode here


Listen to the podcast here


Augment It: Leveraging Data And Artificial Intelligence In Design And Construction With Mehdi Nourbakhsh, Ph.D.

Full disclosure time. I’m ignorant about Artificial Intelligence or AI but I know it’s out there. I know the machines are learning, and I’m pretty confident that we can leverage AI in construction to have a tremendous positive impact. That’s why I’m very excited to have Mehdi Nourbakhsh as a guest. He is an expert in the field of AI. He is the author of Augment It: How Architecture, Engineering, and Construction Leaders Leverage Data and Artificial Intelligence to Build a Sustainable Future.Mehdi defines AI for us on the show. He gives us some use cases for it in construction and paints a clear picture of how construction companies can leverage it to solve tough challenges, get the most out of their employees, and multiply the impact of their best employees across their organization, as well as bottling their secret sauce and using that bottle to increase the effectiveness of their organization. Like I always say, enjoy my conversation with Mehdi. Feel free to share it with other people who you think will benefit from the discussion. Thanks very much for your time, and enjoy this episode.

Mehdi, welcome to the show.

Thanks for having me.

It’s my pleasure. You and I share something in common. For me, it may be a little less apparent than for you but we are both first-generation immigrants to the United States. I’m originally from England. You are from Iran. Tell us about your journey to the United States.

I started my career in structural engineering. As a structural engineer, I started designing buildings and writing codes to design buildings. At some point, I started going into construction and worked as a Construction Manager. That was the first time that I saw firsthand some of the challenges that exist in construction like cost overruns, managing schedules, and lack of laborers. I fell in love with working with people. I started to get a formal education in Construction Management. I had a hunch that technology could help us overcome some of those challenges.

That’s why I did a Master’s in Construction Management in Malaysia. This is where I learned more about construction management in another country and another part of the world. I did a lot of research and development for the Malaysian construction industry, developing constructability standards for the country and being part of an amazing research team there.

From there, I moved to Atlanta, Georgia, doing my PhD in Building Construction and a project in design computing, learning more about architecture and design and how we can bring all of that constructability and requirements from downstream to upstream. This was in Georgia Tech. This was a time that I got introduced to artificial intelligence and learned more about how it can solve some of our problems in design and construction and bridge the gap between these two different worlds.

We are going to dive deeply into artificial intelligence later on in our discussion. I want to begin by asking you this. You started your construction career in Iran, went to Malaysia, and then went to the United States. As you think about those three completely different countries and backgrounds, what differences and similarities did you see in the way that construction gets done in those three countries? Can you give us a little bit on that?

There are lots of similarities. The challenges that we are facing are very similar across the world, such as lack of labor. We had to train farmers to work in construction. I remember days when I used to lose my workers to farmers because they went to farming, and it was a safer environment for them to work in and make more money, perhaps.

These are challenges that exist across the globe in construction. It’s the same as Malaysia and the US but there are different flavors to it in different parts of the world. Cost overruns, delays, you name it. In big projects, these are bigger numbers, and in small projects, these are smaller numbers. They are the same.

As you were getting your PhD, it sounds like you would already have time in the field with hands-on experience. Am I hearing that right as a project engineer?



A technology built right solves a major problem and has at least 10X the return on investment. Click To Tweet


That must have been very useful in terms of being able to tie in the theoretical with the practical.

I used that in my PhD. For that, I worked with Arkansas. Here’s one of the challenges that they had. The building was a hotel. The owner wanted to identify what is the best material for the structure of the building. The owner gave this team 6 months to look at 9 different variations of the design and the material. Is it going to be steel, concrete or timber for this design? They have 3 different designs and 9 combinations. The owner gave them six months to do this analysis because it was very important for the owner. What the team did was it took them one month to come up with a structural design of one building and another month to do the estimating.

Two months passed, and they could do only 1 option out of 9. They said, “This is impossible that we can study all these nine different variations in such a short time.” In my PhD, I developed an AI solution that could quickly, based on the past estimating data and the different data coming from the social invest, that could quickly in a day tell them, “These are the best options for these nine different combinations. This is the solution that is the most economical one.” This way, they could put that in front of the clients instead of 9 months or 6 months in a day or two or a week. This is how they could impress their clients and do their work better and faster, and help the client make better decisions.

Tell me a little bit more about that. As you were presented with that problem, what led you to the solution that you chose?

Often, the problem we are having in our industry is that we do things in sequential order, “First, I need to do the structural design part,” and that takes a month. When the structure is ready, somebody needs to do quantity takeoff and match that with the cost model that we have, “Here is the end result. This is a number and the cost for this building.” The question I always ask is, “Can we do this differently? What if there was a way that we don’t need to do this in a sequential way? What if we can revert that?” Oftentimes, with the help of technology, if it’s built right, we can do that. We can do things faster, cheaper, and in a very short time.

That’s interesting because the key phrase there is, “If it’s built right.” Let’s talk about that a little bit. Your background in construction, how does that help you in understanding the way to build the technology?

This is very important because I often say the technology that is built right solves a major problem. It has at least ten times the return on investment. Working in construction helps me understand what the real problem is, not the problem on the surface but three problems below the problem that is being presented. That’s also matching with what is the right approach and what the right technology is. Even if they are the solution, what is the right technique in AI that can help this company to do things better, faster, to reduce risk, and do that in the most efficient way? Understanding a problem is the key to solving a problem the right way. My background in the industry brings that to the table.

You said something interesting here, and I want to explore that a little bit more. You talked about the problem on the surface and then being able to dive below the presenting problem to get to the root issue. Please give us some insights and the techniques you use to go from the surface problem to the root problem.

One of the well-known techniques in this area is asking five why questions. That’s a very common technique. It’s an awkward exercise often. When you ask that for the 1st, 2nd, and 3rd time, often you hear we are getting closer to the roots of the problem. An example of that was the previous project that I was involved in. The CEO wanted to grow the company. They wanted to increase the revenue of the company without major problems because increasing revenue means hiring more people. How can we do that?

When you ask, “Why do you want to do that?” the answer is, “I want to create more value for the investors to invest in the company and the stakeholders.” When you go deeper into that, you get into what’s the problem five layers below. For this company, that problem was the bidding process. They used to spend a lot of time, maybe two weeks to a month to bid on the project that they couldn’t win or they could win 1 out of 10 or 15 projects that they bid on.

Looking deeper into that, we realized that, going back into the sequential process they have, first, the sales team gets interested in a project, and then their design team needs to do the design. After doing that, the estimating team needs to do the quantity takeoff. Right after that, they realized that, “This is above the budget that the client has. We need to redo the design and the estimating.”

In the end, they had to go back and forth multiple times to come up with a number that made sense. They put that into the bid and don’t get it. That brought $2 million, to be exact, of just waste in the process. We created a technology for them, so instead of a month spending on that, they could do it in a day or two. This way, they could bid on more projects and do that in a very short period.

Let’s pivot a little bit to AI, Artificial Intelligence, because I know that is your specialty. That’s something that you’ve spent a lot of time thinking through and working on. Kick us off here. What is AI? Give us your definition.


COGE 185 | Artificial Intelligence
Artificial Intelligence: You pay for it once and have it for the rest of your life in the company. It won’t leave your company. It will be there to stay, and it doesn’t need that much maintenance.


In one sentence, AI is a field of science where scientists try to design computer systems that mimic human capabilities. We use our eyes to see things. We use our ears to hear things. Our brain does reasoning. We learn, act, and all these, scientists use computer vision techniques to mimic what humans see. Scientists use natural language processing techniques to mimic how human talks. In robotics, they mimic how human acts.

The simple metaphor that I used for people to remember this is that AI is like a tree. It has roots in science, biology, mathematics, computation, evolution, and neuroscience. These are the roots of AI. The branches of AI or different soft fields or techniques in AI could be computer vision, natural language processing, knowledge, representation, reasoning optimization or robotics. Machine learning is another branch of AI. It’s because of the popularity of machine learning and, more specifically deep learning branch of machine learning, people often use machine learning and AI interchangeably. The reality is that machine learning is just one branch of this AI tree.

Tell us what machine learning is.

Machine learning is a technique in AI where scientists try to mimic how humans learn. If you think about humans, we use data every day. We have sensors like our eyes, ears, and tongues. These are all the sensors in our bodies. We capture the data around us. Based on what we capture and how we process it, we learn and connect things to what we know before, and this becomes our knowledge and so on.

Scientist start to mimic that. They say, “What if we have data and lots of data? What if we can learn from this data?” They invented techniques that you provide them with this set of data. These techniques try to learn the relationship or pattern in the data. There are different techniques in machine learning that can do that. Based on those, we need different amounts of data. This is a technique that, if we use, we can learn the pattern of things, recognize objects and use it to act in the environment as some of the robotics. The robots use machine learning techniques to perform certain tasks.

I like that definition of AI looking to mimic human capabilities. It’s very interesting. I would like to then ask, “Why should I, a construction company, president or owner, invest in AI?”

Your reflection was on how AI can mimic human capabilities. What if I tell you that you can mimic your top performer employee capabilities and have that for the rest of the lifetime of your company? Let’s call it a virtual employee that can learn from not only one project but all of your projects and can make suggestions like an adviser can recommend and act on things.

You pay for it once and have it for the rest of your life in the company. It won’t leave your company. It is going to be there to stay. It doesn’t need that much maintenance. That’s the definition of an AI system or AI agent that can mimic human capabilities and help you remove some of the inefficiencies in your company. If you think about lessons learned that comes out of a project, you can bring that up into a scale across your company so that you don’t make the same mistake many times in different projects.

You’ve made a big promise. Your promise is, “I take my A players and replicate them using AI. I don’t even care if my A players get stolen by the competition because I’ve got the AI to help me repeat that excellence on an ongoing basis.” I want to make sure I heard that right.

Yes. AI is like a magic wand that you have. At the end of the day, it’s a tool. It goes back into the problems that you have in the company. What are the major problems that you are facing in your company? If you don’t have any problem, good for you but what are the directions that you are taking your company? It starts with the business strategy. What’s the strategy that you have in the market? How can this tool help you to achieve that strategy?

The example I gave about increasing sales without increasing the headcount is one example of how AI can help you grow your business without increasing your overhead too much. At the end of the day, it’s the tool that can enable you to become more productive, reduce the risk in your projects, and help you make things better and faster.

How do you distinguish between AI and regular software?

The question is, maybe a difference between automation and AI. In automation, we have been automating things for many years. We try to do things faster with automation. The layer that AI ends is learning and being able to get smarter over time. Now, you are using perhaps a CAT tool. The computer-aided design has been here for many years and can help you automate certain tasks but is that learning anything new? Every time you start, you start from a blank screen.


Understanding a problem is the key to solving it the right way. Click To Tweet


The AI layer on top of that could be, for instance, you’ve done a typical project 100 times before. What if you could put in certain parameters of the project like, “This is a hospital?” You have designed a hundred hospitals before, “These are certain parameters. This is the location, and this is what we want to do.” This AI engine can give you a starting point.

Based on what you’ve done in the past, this is your starting point for your design or based on these are the issues and risks that are involved in this specific project in this location, “Watch out for this and that.” This way, you can manage your risk or design better, and you can do your work in a better way. For instance, when an employee comes, you preserve that knowledge in your organization. This AI is able to look at all the data in the past and help your new employees do a better design, manage the risk, and do the project better.

In your mind, what are the limits of AI? The first thing that’s coming into my mind is the word intuition. If I’m looking to mimic human behavior, one of the aspects of human behavior that’s essential but, in many ways, difficult to measure is this idea of intuition.

In writing my book, Augment It, I interviewed and worked with more than 50 people from the industry. I asked them this question, “What can I put in the book that makes it a valuable lead?” I gathered all the information, questions, and things that they had. One of the questions that popped up and it was a pattern, “What AI cannot do and what are the limitations of AI?” I’m happy that you are bringing this up.

One of the things that I often say and also documented in the book is AI cannot imagine new things. Here’s the big picture. There are two ways you can create an AI system. One is you use human knowledge to model things. It’s a modeling-based way of doing this, and/or if you don’t have that knowledge, you have lots of examples from past projects. You provide that. This way, machine learning can learn the patterns and the relationship between all the elements of the data that you provided. These are two different ways.

When you talk about patterns, that sounds a little bit like intuition because that’s part of intuition. It’s able to recognize patterns and process them rapidly in making decisions. Are you saying that AI does have the capability of mimicking intuition?

At the end of the day, it’s what it does, mimicking intuition, yes. If you look at it at a deeper level, it finds the relationship in data. It’s a correlation in data. If you provide the data that has no relationship in it, it gives you bad results because it can’t find any relationship in your data.

Give me an example from the construction of data that are related that could be put into an AI system.

Let’s say safety. Let’s pick that as an example. In safety, for instance, let’s say there is a big relationship between the safety of people and whether or not they have safety equipments on PPE. What you can do as an example is you can create lots of images. You can put cameras on the site and record people as they are moving on the site and then start labeling which has the vest on, who doesn’t have it, who has the helmet on or who doesn’t have one. You can use this data to teach or train a model that can tell now on its own like, “This person has one or doesn’t have one.”

You could use that on the site so that when people are moving around without any personal protective equipment, it tells them, “This person doesn’t.” What you do now is that you have it on all of your cameras on the site. You can immediately say, “This person in this location forgot to put the hard hat on.” They send somebody to inform them or, “This person does it really well, and we need to encourage them.”

This is an example of data that you can provide. To do that, you need to provide data and label them, which means that you need to create a unit copy image and say, “This person has one. This person doesn’t have one.” You need to have enough examples like this. This AI system will learn over time that, “This means hard hat.” It doesn’t really know what a hard hat is but learns over time that, “This is something that this person is interested in. This is the vest.” Now that it has been trained, you can deploy it in all your projects across the globe. You can monitor people if that’s something you want to do.

There are multiple applications here, so let’s be directive here. How should I get started then on my AI journey?

The starting point for executives is I always recommend looking into their business strategies. If the strategy is to offense, to go to new markets, let’s say you want to work on industrial construction. This is a new market for you. You haven’t been there but you want to do that. You want to do it in the next two years and invest in that. This is an area where you can say, “How can I leverage AI to be able to help me to get there faster and better?” There’s always a solution for that.


COGE 185 | Artificial Intelligence
Artificial Intelligence: At the end of the day, AI is the tool that can enable you to become more productive, reduce the risk in your projects, and help you make things better and faster.


If your strategy is continuous improvements, maintaining what you have or defending your position in the market, you can look into your current business problems you have. Is it a cost overrun, and you want to reduce the risk? Whatever problems you are having, look into the opportunities of using AI there. I often say, “Bring your biggest business problems.” What I can help with is working with these executives to say, “These are three different ways that you can solve this problem with the help of technology. This way requires more work. This requires less work. This brings the highest return on investment.”

We typically get into a strategy workshop and bring all those discussions to a table. At the end of the day, we want to find areas with a high return on investment, something that has the most return of value and is easy to do. The result is certain, and we can implement that in a short time. That’s what I recommend to the business executives in the companies.

There are three big buckets here that I’m learning from you. Help me with this. For the first one, I could use AI in terms of design. I could use AI in terms of sales and business growth. I could use AI in terms of quality control or the safety example that you gave on the job site itself.

Another one could be an inspection. Teams at Haskell had only two certified welders. They are responsible for looking at all the weldings and making sure that these are in good shape. They are in the food and beverage industry. For them, it’s very important that the weldings are very clean and certified. They have 100 projects in the US, and these two people can be everywhere. This is challenging. What they did was look into how they could use AI in the hands of the welders. Every welder, when they are done, they take a photo of the welds.

With this photo, the AI system can tell them, “This is a good weld. This is not a good weld.” The certified welders get all this information in their office. They look at them and say, “It is a good weld. This is not a good weld.” This virtual certified welder is in the hands of every welder on the side. This is how they can reduce the risk of having bad welds and lawsuits down the road. It never happened to them. At the end of the day, they are collecting better data. These virtual certified welder is getting better. This is how they are reducing the risk on the site.

What are the biggest mistakes people make when investing in AI?

One is there’s a perception in the industry that for us to get started, we need to have data. To have good data, we need to work for 2 or 3 years to collect that. We are not going to start until 2 or 3 years from now. I had a conversation with one of the executives. He came to my book launch and said, “When you told me that this is not right, I was relieved because we were investing in collecting data for two years.”

This welding example that I gave you is a good example because they said, “We have no example of bad welds or good welds but we want to start in this project. What should we do?” Who keeps examples of 1,000 bad welds? We didn’t want to wait for three years to do that. We can create a data set for good and bad welds based on the limited number of good and bad welds. What they did was they went to the site and asked the welders, “Can you make some bad welds?” They look at them and say, “What is wrong with you? Nobody asks about bad welds.” They were like, “Trust us. Make some bad welds.”

They took photos and brought that back to the office. They created a 3D model of the weld and used that. They were able to create lots of different variations of that weld. This is a starting point or bootstrapping. Over time, as the welders are using the system, they can send the real data back to the office, and we can replace the real data with the data we created. This is how over time, the system gets better and better. As I told you, this is having your best employee mimicking your best employee but over time, it gets better and better. It doesn’t require that much maintenance and can be in every project that you have.

As with all technology, the most important thing when you are first starting out is to create or identify an area that you want to grow in, or an issue or a problem that you have that you want to solve, and be laser-focused on that one thing, to begin with. I want to make sure I hear you correctly.

Yes. Correct.

You have a book that you’ve just published that gets into all of this in detail. Can you tell us about your book, Mehdi?

The book is called Augment It. The reason I wrote this book is that over the past years, I have been working with a lot of different companies in architecture, engineering, construction industry, and manufacturing. I started seeing a pattern of why some of the projects are successful and why some of the projects are not successful. I started documenting those, and now it’s this book.


If we could reduce our risk and do the project better with higher quality, all of us on the planet could benefit from it. It's a win-win for everybody, the industry, and the world. Click To Tweet


I wanted to help the industry avoid some of the mistakes others have been making, sharing my knowledge and how we can work together to push this industry forward. At the end of the day, if you could reduce your risk and do the project better with higher quality, all of us on the planet can benefit from it. It’s a win-win for everybody, the industry, and the world.

If I go out and buy the book, what are the big takeaway or return on my investment of $20 and the time to read it? What am I going to get out of the book?

The first part of the book talks about AI education. It’s about what is AI, what is not, what it can do, what it cannot do, and what are the biases in AI systems. It gives lots of examples in architecture, engineering, and construction, and what are the AI systems in architecture, engineering, and construction. It also talks about the different interactions that you have with AI systems and how you interact and should interact with the AI systems.

This is the first part of the book, how you can use your data, how you can use that in a better and smarter way, some of the challenges you may have, and how you can overcome that. The second part of the book is a how-to book. It talks about now that you know about AI, how can you bring that into your company? When I talk about AI investment, I don’t mean these companies go to create the next flying robots. Let the technology company do that for you.

What I mean here is that you look into the problems that are unique to you or the areas that you are good at. You have a secret sauce in estimating, and you are you because of this. You want to take that to the next level in your company. You spend 10,000 hours or more to do the way you do the estimate, and you want to take that to the next level. This is the area that you need to invest in more, and AI can help you with that.

The second part of the book talks about how you need to do that. It starts with your business strategy. It talks about how you can find these use cases from the different corners of your business, how you should prioritize them, and find the one that has the highest return on investment for you based on your strategy. It talks about different steps you need to take as a leader in your company.

The book is written for ace executives. It’s not for computer scientists. It’s not for a low-level manager. It’s for ace executives and helps them go through the innovation process, how you first need to validate the technical feasibility, the business viability of that idea, and how you can bring that up to your company. You invent, innovate, and augment your business capabilities with AI.

You repeat this process because now that you’ve done it once, you take your company to the next level, and then you can repeat it to go further in the process. That’s the book in a nutshell. The last part of the book brings more examples of actual use cases and case studies of AI in the industry. It brings three examples in design, prefabrication, and construction. In essence, that’s how the book is structured.

It’s a little bit of a twist on your typical approach to technology because when I talk to technology executives who are servicing construction companies, the main thing is, “What problems can we solve?” What I’ve gotten from our discussion here that is interesting to me and that AI offers is this idea of taking your best employee and mimicking and multiplying their effect using AI across your organization.

The second thing you said, which I thought was interesting, is taking your secret sauce, the thing that sets you apart, whether it be an estimating, planning or production, and then replicating and getting that be multiplied across the entire organization and even taking that secret sauce and perhaps executing it quicker and more efficiently.

Also, keep that organizational knowledge in your organization. This is another thing. As a person who works in construction, there are certain key people that, if you lose, you are going to be a big loss for your company and your organization. They get to the retirement stage and are going to leave at some point. It’s being able to have that organizational knowledge in your company. When a junior person comes in, this can help them get up to speed faster and better. This is another angle that we can have.

You could use AI as a mentor for your new employees.

Absolutely. This is how they can get better. Once they acquire more knowledge, they can also contribute to your AI. This is how you improve your organizational knowledge as more data comes in.

COGE 185 | Artificial Intelligence
Augment It: How Architecture, Engineering and Construction Leaders Leverage Data and Artificial Intelligence to Build a Sustainable Future

You have been very generous with your time here. How can people get a copy of your new book, Augment It?

The book is available on Amazon and major booksellers. They can go to, or they can come to my website, There are lots of free resources available. I also have the free AI Workbook available if you want to get started in your AI journey in your company. Go there, click on the Free stuff, download that workbook, and let’s start your journey.

You can go to that website,, and get free book chapters. That’s pretty cool as well. You can check out the book there. Mehdi, you have been very generous. You are in the Bay Area and originally from Iran. If I’m in the Bay Area and hitting a Persian restaurant, what’s the Persian restaurant that I need to go to in the Bay Area?

There are multiple Persian restaurants there. I have a list.

Give us one of your favorites.

My favorite is Lavash. That’s in San Francisco.

Mehdi, you have been very generous with your time. I appreciate it. I’ve learned a lot here about AI. I appreciate your perspective on it. It’s not just about solving problems but also about leveraging your best employees and the best things you do in your organization.

Thanks for having me. Thanks for this great conversation.

Thank you for reading this episode, and I hope you have a better sense of how AI Artificial Intelligence can impact your business positively. Thanks again for reading. If you like the show, and I know many of you do because you are reading, feel free to give us a rating or a review wherever you get your podcasts. That helps us to spread across the interwebs. The more ratings and reviews we get, the more visible the show becomes, and the more construction leaders can benefit from reading. Thanks again. I will catch you on the next episode.


 Important Links


About Mehdi Nourbakhsh

COGE 185 | Artificial IntelligenceMehdi is an author, speaker, and the CEO of YegaTech, a technology consulting company in AEC. He’s devoted to helping CEOs, CTOs, innovation directors, and business executives to grow their businesses and differentiate themselves and their companies via AI technology and innovation.

With a decade of experience in the research and development of innovative AI solutions in the AEC and manufacturing industry at YegaTech, Autodesk, and Georgia Institute of Technology, Mehdi brings a unique perspective to the AEC industry.

Mehdi has led and advised on several AI solutions that are used by tens of thousands of AEC and manufacturing professionals every day.