The AI Revolution in Mechanical Design is Happening

Maorr Farid, co-founder and CEO of Leo AI, shares how his company is revolutionizing mechanical engineering by creating the first AI that truly understands CAD. His mission is to transform engineers from "glorified secretaries" back into creative innovators by eliminating the tedious 85% of time they spend not actually designing.

• Former Unit 8200 intelligence officer with a PhD from Technion at an unprecedented young age
• Founded Leo AI after discovering engineers spend only 15% of their time moving the mouse in CAD software
• Already reached 55,000+ users with zero marketing budget, including major clients like HP, Intel, and Scania
• Leo AI focuses on augmenting engineers rather than replacing them
• Customer testimonials include completing designs in 1.5 hours that previously took three engineers three weeks
• Leo connects to PLMs, Windows directories, and CAD files to leverage organizational knowledge
• Engineers maintain control of decision-making while AI handles information retrieval
• Leo is the only AI system specifically built to understand CAD geometry, not just text
• 80% of elite engineering students leave the profession due to the gap between expectations and reality
• Join the Mechanical Intelligence community - the first global community for AI in mechanical engineering

Transcript

Introduction to Mayar Farid

Roopinder 0:00

Hello and welcome to FoDES the Future of Design and Engineering Software Podcast. My name is Roopinder Tara. On the show we will have guests that will discuss tools and technology that engineers will find interesting and useful. On today's show, I have the pleasure of introducing Maor Farid, co-founder and CEO of Leo AI. Maor, at the age of 33, already has a list of accomplishments a mile long, so we have time to actually talk to him. I've left a few out. He has a Wikipedia page, though, if you want all the details. Most recently, he founded Leo AI along with his classmate at Technion, neighbor and close friend, Moti Moravia. He has managed to round up almost $10 million in funding. Maor was the youngest person ever awarded a PhD at Technion. He was a Fulbright Scholar and did postdoctoral work at MIT. He founded Learn to Succeed and wrote a book of the same name to motivate poor Israeli youth. He was named Forbes 30 Under 30 in 2019.

Roopinder 1:00

He served as a commander in his country's version of the NSA Unit 8200. We join Maor, as he tries to avoid talking about what he did in the 8200. Everybody in Israel has to do military service, but you joined this elite Unit 8200. What is that? It's an elite intelligence unit. I see. S o you probably can't talk about anything you did there. Am I correct?

Roopinder 1:24

I see so you probably can't talk about anything you did there, am I correct?

Maor 1:27

Yeah, just that the food was great

Roopinder 1:28

Was it?

Maor 1:29

Not really.

Maor 1:31

I liked it.

Roopinder 1:32

Did you use AI there? We developed algorithms that identify targets for homeland security purposes.

Maor 1:48

Some of the things I hear about Israeli intelligence, amazing technology, the plots that have come out of there, the pagers. I was not working on it, and if I worked on it, this is exactly what I would say I haven't worked on it, but I really didn't work on it.

Roopinder 1:56

Either way, the technology is admirable. I find this problem all the time. Tell me if this happens to you. When I'm traveling abroad, I often have to explain or sometimes apologize for what our leaders are doing. People say, oh, what is this president of yours? And this has happened not just with our president, but previous presidents as well oh, what is your president doing to us? Are you having to do that? You just came to the US recently. Are you encountering that kind of thing with what's going on?

Maor 2:26

This is a very interesting question because when I was at MIT, I faced not only prejudice based on my Israeliness, but people didn't really know what my background was not because I have a heavy accent. They didn't know that if I'm Muslim or Jewish, because my ancestors came from Iraq and Libya, so we're a very unique breed or community. It's called Sephardic Jews, Jews that came from Arab countries, so some people thought that I'm Muslim, so they had their own prejudice about Islam, and also when people knew that I'm Jewish, so I'm somewhere in the middle. But I don't feel the urge to apologize Because I know exactly what I've done in my military service. I'm really proud of what I did and what my friends did and this is a hundred percent protecting home and our families. If you know there is a leader or leaders or people that have different agendas or something and I'm not talking about specific person I don't feel the urge to apologize on their behalf.

Roopinder 3:41

Fair enough, we have this podcast and it sounds like you've already listened to some of it. We've had other people on. Jon Hirschtick was recorded just a couple of weeks ago, so you're in good company.

Maor 3:53

Just to hear a friend of Jon Hirschtick, it gives me chills and I'll tell you maybe, if you'd like, in the next episode, why I'm so excited. He had a very dramatic impact on my career, and also Scott and the SolidWorks gang.

Roopinder 4:08

Go ahead, you have time. As long as you have time, tell me your realistic story.

Maor 4:11

So you know, I dreamt to become a mechanical engineer in July 12, 2006. Vacation with my family this is the day when the Israel-Lebanon war started. We were bombarded and we fled out of there, and I dreamt to solve this problem one day, to protect home, et cetera. I realized that the way to do it is to develop technology. This is when my dream to become an engineer started, when I was in 10th grade. By mistake, I stumbled upon SolidWorks when I was in high school during the breaks. The moment I touched the mouse, I was hooked. Then I decided that I'm going to become a mechanical engineer. From this moment on, the bachelor's, master's, PhD and the postdoc were in mechanical engineering, and it wouldn't be possible without SolidWorks, Jon, Scott and the magnificent people in the SolidWorks gang.

Roopinder 5:06

I'm always interested in how people got their ideas and what made them go into this field. Yours is much more inspirational. I think I stumbled on publishing accidentally, but I think of myself as a mechanical engineer first. We mechanical engineers and I shared that with J Hirschtick too. It's like we have this special bond like secret handshake or call it what you will. We get each other. Yeah, so I understand. Your relationship with Hirschtik goes a little bit deeper. I got a message after I signed up for Leo, your salesperson, and is this somebody? Hirschtick? I thought, hey, that's an interesting name. Are you in any way related to John Hirschtick? And he goes 'that's my dad.' How did that happen? Is there more to this investment by John Hirschtick than giving you a son? What?

Maor 5:53

I could share here is that we were looking for a stallion SDR and, first of all, you know, we started a LinkedIn post, I mean like a job offer, and then something like don't catch me exactly on the number, but something like 500 people applied. It's super noisy and we started to activate our first circle of people that we really trust and they know one or two people around. John is one of them. He shared with us about his son, Sam. He said meet him and see. What do you think? And I'm telling you, Roopinder, they're just between us. He's a 10x person, super intelligent, super motivated.

Journey to Mechanical Engineering

Roopinder 6:38

I imagine he gets quite a bit of that from his dad. Has Hirschtick invested in Leo?

Roopinder 6:45

We call this a happy hour. You're in Israel, what time. The idea of happy hour started because I meet interesting people and then we got to talk about talk like human beings, person to person, rather than about work. We always end up talking about work anyway, because anybody I know who's passionate work. It's not work to them, it's their life.

Roopinder 7:04

Mission and obsession you started, leo. That's your obsession, if I can call it that, and that's AI CAD. How has this been received? Tell me about how it's doing.

Maor 7:15

Long story short. We started in May 2023. With zero dollars in marketing, we got to more than 55,000 all-time users, customers like HP, Scania, Intel, Mobila a very long list. And forget about logos. The things that makes our heart beat and what we read on our Thursday happy hour, with all of our team around the table eating chocolate and stuff, is what our customers are saying. We had one customer that I could share because he was joining our last webinar and he said that, using Leo, he was able to design a system by himself in one and a half hours that he used to design similar order of magnitude of complexity with three peers, three weeks. He has 20 years of experience in the defense and aerospace industry, so he's not the junior kind of guy. This is what makes us wake up in the morning.

Roopinder 8:17

You've started other companies before and they're not always been about computer-aided design, right? What made you think, okay, CAD is now ready for AI. What was going through your head, all of us seeing all this stuff about AI? And I'll tell you what happened to me. I was frustrated. The CAD companies were not putting AI into their product and I talked to every major CAD company and said, hey look, we need AI.

Roopinder 8:45

Can you make CAD smart enough to do these things? I've always thought it should, but it never has been able to. I always thought CAD is not computer aided at all. That's a fallacy or a misrepresentation. Computer aided design was never computer aided. Looking back on it, all it was. You just find a way to document. You're physically describing a part. That's it. I'm designing in my head and I'm asking the CAD product to give it a geometric form Computer represented geometry. Maybe the things that the CAD industry has tried to pass off as intelligence has been topology optimization, which I don't consider very smart at all. I don't think it's giving us good shape. So what is Leo bringing to the table that's actually making CAD smart?

Maor 9:37

I'll start with addressing the first part of your statement, when I thought that applying AI to CAD is going to be a good idea. But the reality is it was never about CAD. CAD is not design. CAD is not mechanical engineering. It's not engineering, it's a tool. It's like cooking and talking about pans. We're going to SolidWorks or OnShape only when we know exactly what we want to design. I talked with one of the OGs that I'm very happy to talk with, Bertrand Sicot, the former CEO of SolidW orks and a mentor for us, and he said that when he was the CEO of solid works, he was going through the usage data and he said that only 15 percent of the time, SolidWorks that is open, the engineer literally moves the mouse. What happens in the 85? This is what led us to start, Leo.

Founding Leo AI and Early Success

Maor 10:26

I told you that I joined the Israeli Defense Forces and the elite unit that trains mechanical engineers it's called Brakim program and I did my bachelor's, master's and also the PhD. It was like a honeymoon the academic version of honeymoon. I learned about thermodynamics and Newton's laws. I felt that I understand nature, how the world really works. I dreamt on this day that I'm going to become a mechanical engineer and I'm going to do this magic turn all those principles into robots that make our life safer and medical devices are going to make us live longer. And then the day has come and I started to work as a mechanical engineer and Roopinder, it was the worst days of my life. I remember the first three months. I was spending all of those days trying to find three bolts and two bolt bearings. None of them even were a good fit. I realized that instead of innovating and solving problems, I was a glorified secretary. I was going through this endless PDF and vendor catalogs and the PLM I think 1% of the time I was using SolidWorks and then I realized that something is wrong. Some of the first memories that we have from our childhood it's when we broke things apart, trying to understand how things are working, trying to build stuff from cubes, wooden cubes. It's almost a human instinct to build, to design, to innovate, to create. Unfortunately, mechanical engineering is not there. To innovate, to create. Unfortunately, mechanical engineering is not there.

Maor 11:56

Millions of talented people spend their time on finding information in PDF documents, in CAD, in PLMs, and it's not really working. It's a knowledge problem. Engineering knowledge is still siloed in the minds of the older jobs at the end of the corridor, the more experienced engineer that are gonna retire soon, the documents that the organization wrote in the last two decades, the CAD experienced engineers that are going to retire soon, the documents that the organization wrote in the last two decades, the CAD files that they designed and the Bibles of engineering. So with that notion, moti and I, when we started Leo, we wanted to understand where to start from. What is the AI for engineers? Because we know how AI for lawyers look like it's called Harvey and we know how AI for developers looks like it's called GitHub, copilot. How the AI for engineers looks like.

Maor 12:47

So Hollywood says that it's Iron man or Jarvis. Right, you are talking to a machine and then it does the CAD. It's BS. We realize that no engineer is going to use it in the next two decades. Okay, we won't trust an agent that will create all the CAD for us. We want to do the CAD. We love it. Just save me, not the 20% that I love which is building, doing CAD. Save me the 80% of me going to the PLM, writing documents, changing them, talking to peers. Right, we surveyed and interviewed more than 900 mechanical engineers, from juniors to VP of engineering, and we started from the most time-consuming tasks that the engineers are doing today. This is what Leo does and this is how we save on average five hours of tedious work per engineer per week.

The Knowledge Problem in Engineering

Roopinder 13:39

That's very interesting. You reminded me of something I just read. I was doing research on AI and how it's affecting people and I turned up a study done by an employment agency that said mechanical engineers are at the second or third in the most disillusioned disappointed people as they get into their career. Earlier in their career. Like you got into engineering expecting to do great things, you ended up trying to find parts in a PLM system. I had a very similar experience 45 years ago when I got my first engineering job. Like you, I wanted to be an engineer and make great things.

Roopinder 14:20

Since I was a kid, I envisioned myself making beautiful planes. I would be sketching airplanes as a child and I ended up making drawings by hand of 3D parts, and every day I made a drawing on paper. This was a six-month co-op session. Every day I made one drawing. I was so bored I almost left the profession, but the funny thing is I look back on it. AI still can't make 2D drawings. AI is not doing what engineers want to do. This is why I'm very hopeful Leo is going to actually help us. I'm running out of time, but I think for the next generation, Leo could be that company that saves engineers from having to do this drudge work. What do you envision Leo doing? Do you envision that it'll displace CAD, or do you think it'll be an add-on to CAD? If I want to say, make me a bicycle, make me a part for an airplane, right, is that going to do that instead of CAD? How's that going to work?

Maor 15:25

Address this question, I will address a tougher question. If Leo is thinking about displacing mechanical engineers and the answer is, god forbid the opposite, we want any mechanical engineer to focus on what they love and do best, which is build beautiful, fantastic products. In my vision, I want to make the correction. What I experienced and what 80% of my class at the Technion experienced.

Maor 15:52

By the way, 80% left mechanical engineering to other professions, 80% from the Brokheim program, this military program experience. By the way, 80 percent left mechanical engineering to other professions, 80 percent from the brooklyn program, this military program that's the highest attrition rate I've ever heard of 80.

Roopinder 16:04

Is that because it was really difficult?

Maor 16:06

No, because it was so depressing, because they were the most talented youth from israel. They want to design robots and eventually they found themselves for years finding ball bearings in hard copy catalogs written in German, and they don't speak German.

Roopinder 16:23

These were the smartest people, maybe in the country. It was not their intelligence, it was other factors limited, prevented from being good engineers by language.

Maor 16:31

The profession was exactly the opposite of what they envisioned. They envisioned innovation, creativity. Eventually, it was paper pushing. We do not want to displace engineers. The opposite we want to make every engineer focus on what they do best and what they love most, which is build, innovate and create, and we want to save this. 80% of them trying to collect information from here and there. It was a strategic question to answer. Are we focused on helping them in the drawings or helping them in the design? So, from our values core values of allowing human engineers to design better products, I think the answer should be help them in the design stage rather than the manufacturing or documentation stage, Because the robot is going to be the same robot if you're going to do all the drawing for him. But if you help them find the best engine, find the best mechanism for the robot or medical device, then you're going to make better medical devices and robots. So this is our DNA, at least.

Roopinder 17:38

What would a prompt look like for Leo? What would I actually be saying? I would not say make me a great robot.

Maor 17:45

An ideal prompt for Leo is I need to design a suspension system for my landing gear. And why is that? We're talking in levels of mechanisms and assemblies? Because this is the work amount that we feel comfortable as engineers to outsource to a junior engineer, which is explainable. I don't need to go through all the documents and decisions. When you bring the best spring, the best mechanism, the best sub-assembly that I could check in a few seconds or minutes. I feel comfortable because one of our customers in a massive automotive conglomerate said Maor, you're working in the decision-making business, not in the automotive business, and mechanical engineer it means that he or she are decision makers. So you need to feel comfortable with your decision. If Leo is going to design a full robot for you, you're going to say I cannot sign on this or you're going to take me ages to find mistakes here. I'm going to turn into a spy or a detective.

Roopinder 18:52

Spoken like a real engineer. I have to say this is music to my ears because I've always said, when I hear about wild claims of AI, like I've heard actual design companies say this, make me a car. I've had AEC companies sometimes in joke but oftentimes not in joke say make me a house. They expect these to be the prompts for their programs. I hear real architects and real engineers saying hey, if that plane is designed by AI, that's not a plane I want to fly on, that's not a building I want to be in if AI designed it. They all say I know how to make these things. I need help to make these things.

Roopinder 19:36

If I'm making a building, I don't want to design the I-beams. I want it to make the I-beams. I want it to find the right I-beams. If I'm making a, I always use this as an example. If I make a bicycle, make it out of tubes. I know tubes work. Tubes are good for bicycles, right. If I'm making a car work, tubes are good for bicycles, right.

Roopinder 19:59

If I'm making a car, don't make me a car. I know how to make cars. Help me electrify a car or help me find a suspension for now that this electric car, the weights are going to be different. Make me a suspension that works. What I wanted to do is solve known problems. Call this conceited, but we, as engineers, think we have the right idea, but we want help in making it, Would you agree? We just don't want to do that grindy work right. We don't want to start from scratch and make blobs that eventually will be beams or rods. We want it to say hey look, I'm making a bicycle, make it out of tubes. I love tubes. Or if I'm making a building, make it out of I-beams I-beams work right. That's something you would believe in.

Leo's Approach to AI for Engineers

Maor 20:43

I believe that, as an engineer myself, I think we should keep the control of the decision-making process. You can think about it. Okay, let's replace the concept. Just provide me concept ideas, parts, whatever, if you're thinking, thinking about it, the jobs that you want ai to do for you or a junior to do for you. What's the difference between them and what you do want to do yourself is one thing in common control, the things that I don't care to lose control. Okay, give me concept, I would choose one. Okay, give me ball bearings, I would choose one. Or do the drawing. Okay, something like, something like this.

Maor 21:17

But I do not blame the players on our market that are talking about hey, AI, design me a car. I think I'm not judging God forbid, it's just my two cents. But I think it's something that we matured into as Leo AI. We thought about generating concepts and ideas and realized there is a very deep objection. We utilized the five whys principle by Toyota. I'm not sure exactly about who invented it in Toyota, but I think one of their leaders in the 80s, when they were losing significant market shares for their competitors, he said let's ask five times why to our customers why they're choosing the competitor, and then five times why. And then the reality. The core reason is going to be visible for us. If you're asking someone, why do you choose Volvo? I love the brand. Why? Because I love the commercial. Why? Because it seems safer? Why? Because I don't want my family to die in a car accident? Okay, so it's about your family, it's about safety.

Maor 22:21

So, when we've done a lot of A-B testing at Leo at the beginning in May 2023, we put a website that turned ideas into products in seconds and we saw many people click on it. And then we interviewed people and realized that there is a very strong objection and after drinking beer with some of them, they told me more. I love your website, I love your state of mind, but I'm terrified to death because I'm going to be unemployed. So it's the maturity of the builder I'm not saying words like entrepreneur, but as a builder, someone that wants to find a solution to a problem is to understand exactly what is the top of letter needs, what are the core needs, fears, concerns of your customer. And put aside all the fairy dust, AI BS things. I'm an AI researcher first and a mechanical engineer. I studied AI before. It was a buzzword and I despise people throwing those two letters on every product. Right, it's an AI bottle, buy it. It's the same bottle. It's an AI toilet. No, if you don't need or want AI, don't use it.

Maor 23:29

I think a good principle is IA rather than AI, which is intelligence augmentation. I think and this is something Scott Harris says rather than AI, which is intelligence augmentation. I think and this is something Scott Harris says like, find a use case, a small use case, that you can design a car for me. It's not doing one thing. It's not designing a car. It's choosing the springs. The mates is writing the document is doing a thousand different things. So I think it's a bad practice, but this is only my approach. Yeah, I think it's a bad practice, but this is only my approach.

Roopinder 23:58

Yeah, I think it's certainly valid. I agree. Autodesk at Autodesk University introduced a Make Me a Car software they just bought because they wanted to show off how they designed a shape of a car. Nothing inside, no suspension, no engine, nothing, just a shape. You could maybe use it for CFD, but that was it, and I was very critical of that concept. I don't want it to design a car. I'm an engineer but I do need help If I want to electrify a car.

Roopinder 24:30

That is a known problem, a known solution. Ford, let's say, has electrified several cars. Mercedes has electrified several cars. That's a known problem. You know how to substitute electric motor for internal combustion motor. You know how the weight is going to shift. You know how the suspension is going to change. You have to do something with the space that you save, right. We have to do wiring and computerizing. You've done this already. You have a database. You have tribal knowledge. Maybe it's only a few cases, but this is a problem you have solved. Now AI should be able to assist you with steps along the way of electrifying your product, because you already have history of it. In the next generation of vehicle design, a prompt could be made that said electrify this and with enough data, it would go out and do this. Give you several options along the way, but seize upon parts you've already created, designs that you've already created, and electrify. The next entire product line is going to switch from IC to electric motors. Do it, Does that sound too far-fetched?

Maor 25:36

No, it's not far-fetched. This is exactly what we're doing at Leo. At Leo, this is our claim to fame leverage your organizational knowledge to design, to make better design decisions faster. Leo is connected into your PLM, into your Windows directory, into your CAD, read all your documents and this is the first AI that understands CAD. This is our proprietary, patented AI model the first to understand CAD. And imagine that it read all your CAD and text files. It knows all your tribal knowledge. It broke all the silos and also it exposed only to the 1 million bibles of engineering, not Reddit posts like ChatGPT, right, and in this way, just the super- experienced engineer behind your shoulder.

Maor 26:26

And when you need the next part, when you need the spring, you just tell Leo I need the spring here, those are the parameters that he needs to, those are the requirements. This load, this compression. Leo tells you what are the parameters that he needs. If you didn't provide them, it runs the calculation based on sources from your data or from the Bibles of engineering and eventually finds you the part that you need engineering and eventually finds you the part that you need. So in this way, we're meeting our vision of your engineer, turning into a manager, into a leader of the design process. You tell the AI hey, I need a vault here. What's the best one? Okay, what's the guidelines say, got it. There's a conflict between ISO meal standard and give me a table. Okay, I made the decision, choose this, put give me a table. Okay, I made a decision, choose this, Put this here. Assemble Okay, I feel comfortable with it.

Roopinder 27:19

Yeah, that has to do a lot with comfort. We want to be still in the driver's seat, but we'd like a little co-pilot next to us, like a navigator we can trust.

Roopinder 27:28

I think this is why we talked about SolidWorks and John Hirschtick and I think that's what helped him be very successful, because he was an engineer. He created a mechanical engineering product, a tool, and engineers trusted him because he was an engineer. Trust is very important. We trust our peers more than we trust developers, who are not engineers. They're just programmers. You touched on a subject that is very important.

Trust and Fear in AI Adoption

Roopinder 28:02

I think people don't talk about it as much as they should, especially people like us who are into technology, and that's how technology can sometimes come at a social cost. You said you went to engineers and your fourth or fifth Why? You said you went to engineers and your fourth or fifth Why? they mentioned, by the way, you might want to change it to four or five beers instead of five Whys, but after a few beers they do break down and tell you they're afraid. I had this happen to me. I was on a panel of AI experts. I don't consider myself one. Eventually, it came out that people were scared of AI. Eventually, it came out that people were scared of AI. When I asked people like who's using AI, two or three hands went up in a room of 30 people. They said I used AI and it came out after the fourth, fifth why they were scared of it. This is a very palpable fear. Engineers, as much as we think about how indispensable we are, we're looking over our shoulders behind us and see AI is learning really, really fast. I don't know if I should tell my kids to go be a mechanical engineer. Right, are they going to be replaced? What do you think? I think to some extent, you can say yeah, it'll make you a much better engineer. You'll have an AI agent with you that's going to help you do your job. How do you counter this fear that eventually it is going to do real engineering or there'll be less of you? There might be a super engineer like you, but will there be enough to support your children who want to be engineers?

Maor 29:32

When the mechanical engineers I drank beers with told me that they are scared, I think I was one beer away from the real answer. Right, it was after four beers they said they're afraid. But at the fifth beer, I think if you would ask them why are you scared, they're going to tell you and that's what they told me. I do not trust AI. There are two reasons. First of all, because they were exposed to CheGPT and they saw that in our field something like give or take. This is the data that we collected and, based on several models of ChatGPT, 46% of the result that ChatGPT provided were hallucinations, specifically for mechanical engineering. So it means that the mechanical engineer that I drank beers with tried ChatGPT on choosing coatings, bolts, bearings, and it hallucinated. And when the engineer pushed ChatGPT to the wall and he or she asked them, are you sure? They said oh, no, you're right, I was mistaken. So this experience made them feel uncomfortable and don't trust AI.

Maor 30:42

And the other thing is xenophobia. Xenophobia, you know, is the fear of foreigners or foreign powers or entities. I think that because mechanical engineering have never and it was before the Leo era, from my experience before they touched or played with Leo, they said, oh, I'm afraid. But when they started to work with Leo, it became their friend or co-worker, co-pilot. They started to trust it. This is why our customers say some of them quote Leo is the only AI that I can trust. And trust, this is the word that they want to maximize. This is what we ask our customers, not how you like the system, how it helps you, how much time is saved. The first thing is in which extent you feel that you can trust Leo.

Roopinder 31:25

It's amazing how quickly it's learning. In the US they have a PE exam and it comes in two parts. One is the fundamentals of engineering, the other one is the PE professional engineering exam. They're both eight hours long. The first part is what I'd say is book learning. It tests you on all the math, science and engineering theory Cornell did a study on. They gave AI ChatGPT version 4. This was two years ago. They gave it the fundamental, the FE test, the fundamental of engineering test. It got 70% correct. It's passing. Great, it can pass the fundamental of engineering exam and I thought, wow, that's something I did flunk the next test, which is the PE exam. Now, the PE exam is different. It's also an eight-hour test. It tests not just your knowledge of fundamentals, it tests your judgment. It doesn't just test your intelligence; it tests your wisdom, shall we say. I understand.

Roopinder 32:34

You can't even take it unless you've been in the field. You have to have some experience. It got 46%. It didn't pass not a passing grade. It's not there yet. But the Cornell researchers said in their summation just give it time. Chat GPT-5 is already out. They said it'll probably pass the PE exam.

Maor 32:52

They said it'll probably pass the PE exam, but I highly disagree, highly disagree, and it's not because I'm short-sighted and saying that, oh, AI would never be so smart, but I think that the 70 plus percent that you mentioned, or the 46, are about to 10% for mechanical engineering, and I will explain. It's all about the modality. And Yann LeCun is a French, one of the AI GOATs, one of the OGs of AI worldwide, and he said something which I highly agree with. He said that language models, large language models, will always be bounded by the problem that they can solve that come from the language domain. So, if the problem in hand is give me the best medicine for this and that disease, it's just about memory and collecting verbal information and wrap it up and give it for you. However, if the modality of the question is not only lingual, you cannot solve the problem.

Maor 33:52

In other terms, you need AI to understand volume. You need AI to understand physics, the rules of physics, not only language. In order to predict the trajectories of planets around the sun and to understand what's the vector field, they need to understand the connection between motion and forces, the underlying laws of physics. They need to understand volumes, but language models it's embedded in their name, they know how to process language. So ChatGPT. If there are going to be language models stronger, bigger, more handsome language models they will be always bounded by the most junior mechanical engineer, that every mechanical engineer should rotate the design and see oh, there is a hole here, because it can digest B-rep geometry. So Yann LeCun says that only when AI is going to be able to do to digest other modalities, they're going to be able to design stuff, et cetera. And this is what we're doing in Leo. This is why we focused as the first AI made by mechanical engineers. The first thing we needed to do is to create an AI model that knows to digest and understand geometry or CAD.

The Modality Challenge in Engineering AI

Roopinder 35:16

That sounds very exciting. This is much needed. I wish you the best of luck. Have you encountered any besides xenophobia and maybe a fear, internal fear of being displaced? Have you encountered any resistance to, except AI or Leo oh?

Maor 35:35

Yeah, before Leo started, we started to see the hockey stick of adoption. Almost today, 90% of the inbound customers or prospects that reach out to us use Leo. They don't have this xenophobia anymore. But before we saw the adoption, people just wanted to learn about it. There was xenophobia. But today the market in another stage of maturity, of adopting AI, they know that it's here to stay and it has a lot of potential in helping them do better and more of what they love to do. It's not replacing CAD design, just helping them do more CAD design rather than writing and reading documents.

Roopinder 36:24

Yeah, all right. This has been gratifying to hear, very encouraging. It's the first time I've talked to anybody on the subject of AI where I felt like this is going to happen in my lifetime. I've been impatiently waiting for AI to come into. I think you're going to make it happen. Best of luck, I'll give you all the support I can Appreciate that. I better let you get on with your evening. What have you got planned? Are you going out for dinner or going home?

Maor 36:55

Oh God. So the day just starts. I have meetings with customers because we're obsessed, right, so we need to make sure that they're happy until 10 pm, my time. Then I'm going to start my shift as a babysitter with our baby two months old baby, david. So my shift is going to start and I'm going to do my real duty, which is being a dad, and this is the most fun part of the day.

Roopinder 37:15

Tiring, but fun. Yeah, excellent. You already look like you're fresh, even though the day's already over for me, but good for you, you've got some stamina. Maor, it's been an excellent talk to you, have a great evening and I look forward to great things.

Maor 37:30

Thank you so much, Roopinder, and I invite you to join the MI community, the mechanical intelligence community that we started, the first global community for AI, for mechanical engineers. All the tools, studies, but only one rule no BS, only real things, real studies, real tools by and for mechanical engineers. So I would love you to be there, share your thoughts with the other mechanical engineers, and vice versa, I'm sure that you're going to enjoy there.

Closing Thoughts and MI Community

Roopinder 38:02

I love it. I will help promote that too. It sounds like a wonderful idea. All right, great to see you again and I look forward to our next time. Thank you for listening to FoDES, the Future of Design and Engineering Software Show brought to you by ENGtechnica, I hope you have learned of a new application or technology that will help you with your job. If you have an application you think would be of interest to other engineers, please let me know by emailing me at roopinder@ engtechnica. com or message me on LinkedIn.