Industry Interview: Julien Moutte: The Successor to Keith Bentley

Julien Moutte (right) was named CTO of Bentley Systems, taking over from longtime CTO and founder, Keith Bentley (left).

---

In December 2023, Informed Infrastructure Editorial Director Todd Danielson interviewed Julien Moutte, the new chief technology officer (CTO) at Bentley Systems. Moutte succeeded Keith Bentley, one of the company’s founders and its only CTO since 1984.

The full video interview can be viewed at bit.ly/3PWPHg3, but we also wanted to include some excerpts in this issue of the print magazine.

---

Danielson: Do you feel pressure being the first official successor to Keith Bentley, who has been there since the company started?

Moutte: Yes, those are big shoes to fill, for sure, but it’s also a very exciting experience for me. I joined Bentley because I was inspired by the vision of Keith Bentley and very aligned with the way he approaches things. We share that same vision: we want to advance how infrastructure is designed, built and operated. And we both believe it can only be done by becoming data centric and using digital twins. I’m planning to continue that journey, carry the flag even further and accelerate the delivery on that vision.

There’s also pressure that the world is changing fast, and we see the emergence of artificial intelligence (AI) or, more-specifically, generative artificial intelligence. We are working and seeing how much of that pressure we can handle to push our products in the right direction and enhance them with those capabilities for the benefits of our users and accounts.

Danielson: Would you say you’re trying to maintain consistency in this role?

Moutte: Absolutely, yes. We are really trying to drive that same vision, which Keith started quite some years ago, about creating that platform for digital twins, which we call the iTwin Platform. It is the foundation that all our products are leveraging to integrate with each other, but also to expose engineering data to our users. That is something we both believe is needed in our industry and that I’m planning to deploy together with the teams in our technology strategy and vision.

 

Danielson: Can you summarize Bentley’s AI philosophy and how the company is adding AI into its product portfolio?

Moutte: It’s important to realize that AI has been a part of our products for quite some time already. There’s a lot of different fields in AI, and one of the fields of AI that we’ve been using quite substantially in our products is computer vision. Being able to analyze photos, videos and point clouds to detect cracks, rust, vegetation and spalling is something our users have been benefiting from for quite some time.

But it is also being able to do predictive maintenance; understanding when something might fail based on the historical data, the readings, the IoT stencils that we can capture. AI has been there for a while, and it’s already delivering a lot of benefits.

Now, we get a lot of questions about generative AI. What are the capabilities for leveraging the existing data of engineering firms to help them create designs faster and automate a lot of the work that they are doing? As generative AI is appearing, many are concerned about what’s going to happen. Is AI going to do the work? In the infrastructure sector, we don’t believe that’s the case. We believe AI will not replace engineers; it will augment engineers with increased productivity and a lot of automation.

Engineers will be able to leverage those AI capabilities—they will work faster, they will be more efficient than other engineers. And that will make, I believe, a difference in the market.

Danielson: What do you say to engineers who think AI is hype and maybe not something they need to jump in with until it’s become more evolved?

Moutte: Speaking about the broad term artificial intelligence—not only generative AI—I don’t think there’s any turning back. All the applications with automated asset inspections with drones and being able to analyze thousands of pictures and point clouds automatically to capture the defects they need to address—they don’t want to have people going back on bridges and hanging from ropes to be able to inspect those assets.

There are also the users and teams that have been able to automate the design of structures and consider thousands of alternatives to find the best possible design. Again, there’s no turning back from them. The productivity gain that they’ve achieved is something they cannot let go.

I think AI is here to stay. It’s not hype. In the engineering world, we have a much better understanding of the fact that it’s not only about statistics and being able to infer the right answer to some specific question. There needs to be some reasoning. And we’ve seen that large-language models have limitations in reasoning as well.

We need to find the right way to leverage those AI models together with the intelligence that is already in those existing applications to provide design recommendations that will help accelerate the work of those engineers. An engineer who can consider a lot more alternatives and explore the design space much faster because AI is able to provide recommendations that are informed by structural analysis, simulations, carbon footprints and you name it—they’re going to be so much more productive. The designs and the infrastructure we built with those applications will be so much better.

Danielson: How do the owners and builders know if the people doing designs are relying on AI too much without that human oversight? Are rules being developed for engineering to limit what’s acceptable use of AI? And does engineering and infrastructure need such rules?

Moutte: I think we do, like many sectors. And it’s inspiring to see that, for instance, the U.S. and EU have been moving very fast—at least much faster than in the past—on developing regulation around AI because they see the urgency.

I think those rules would be beneficial because they are looking at the usage of AI systems, and they are ranking them based on risks and safety. And I think we can agree that infrastructure is going to rank pretty high in that risk and safety scale, and those rules are going to be very welcome.

We shouldn’t let the fear of what AI can bring deter us from being able to harvest the efficiency gains that it can bring to the table. Computers have been helping with design for quite some time already; that was the basis of computer-assisted drawing (CAD). A lot of designs being built today, even though they might not be using generative AI, have been using computer applications to generate a lot of that design and validate the fact that the design is safe and sound, and it’s going to comply with the regulation and codes.

As long as this is enforced, the owners don’t need to be worried about the fact that AI has been used to craft some of those designs and analyze potential. It’s just a different algorithm to arrive to a better result.

Danielson: Another major concern is the intellectual property of users’ data, not just in the projects themselves, but in the AI models. Can you summarize Bentley’s philosophy on intellectual property and users’ data?

Moutte: That’s a really important point for us. The data for our users is their data always, and they retain control on how that data is being used and if it’s being used. That is so important because the ownership of engineering data is sometimes not very well understood. Is the owner of that data the owner of the asset that commanded the work? Is it the design and engineering firm that produced the models where the usage of the data is clearly defined at the beginning of the contract? And can the data be reused for future projects?

There are a lot of questions there, and our users need to have control over how the data is potentially being used. They need to be able to consent to the usage of that data. Then, when it’s being used, they want to make sure that the data is benefiting their colleagues and their work, but they don’t want the data to be used to train AI models that are going to benefit a competitor that is going to beat against them in a specific project. For us, this is clearly a very important point that we’ve made to our teams and which we are enforcing in our applications.

 

Danielson: Can you summarize your thoughts on the automation of drawings and how AI can save a lot of time there?

Moutte: AI is not going to replace engineers, but it’s going to augment them with automation and productivity gains. One of the key pain points that we see and hear from our users and accounts is the fact that we do 3D design and modeling with our applications, but they have to produce drawings on paper to convey that information. And that takes so much time in their day-to-day activities. Between 30 percent to 50 percent of the project cost is consumed by the production of drawings.

Applying AI to help designers generate better designs faster is one thing. But applying AI to delegate or automate some of those repetitive and tedious tasks is also very beneficial, so we can also address some of the talent shortage that the market is experiencing. There’s so much work to be done with the infrastructure deals, with all the climate-change pressure that is imposed on infrastructure. And those engineers, we don’t have enough of them; they don’t have enough time. They’re working more than they used to in the past. And we want to make sure that AI is helping them handle all that work.

The automation of drawing production is clearly a key point that would have a massive impact on their work life. AI is very well suited for that, understanding and detecting what part of the models are relevant to produce a drawing. Because it’s automated, it can almost be done on demand. Making sure those drawings have the information conveyed in the most clear and readable way is also something AI can be really good at.

Danielson: You’re also passionate about using technology to limit climate change and its damage. Can you summarize how you see engineering and Bentley technology aiding in this endeavor?

Moutte: Infrastructure needs to be built differently because we want that infrastructure to be creating a solution and not contributing to the problem. Concrete, the transportation, and the energy to build infrastructure is potentially something that can contribute [to climate change], and we need to build that infrastructure better. This is what we call our handprint, and we try to make sure our applications are able to help engineers and designers create infrastructure in the best possible way so it consumes less material, it is easier to construct, and it has a lower carbon footprint and environmental impact where it’s being built.

We also want infrastructure to last for 100 years in a way that is going to be resilient. Again, AI can help those engineers consider what will be the performance of a specific design when there is a hurricane, scorching summer, a flood or the sea level rising. Technology can help a lot by making sure that we are part of the solution, not the problem, and that we build infrastructure that is going to sustain our quality of life in the most resilient
way possible.