Uzair Sayid of NexCAD - AI Catches Your Drawing Mistakes

We sit down with Uzair Sayid, founder of NextCad AI, to explore how automated drawing checks cut busywork, reduce errors, and capture expert standards without slowing design. The conversation tracks his journey from frustrated mechanical engineer to building an on-prem tool that blends rules with AI to expertly check engineering drawings.

• Origin story rooted in wasted time on documentation
• Local, secure checker for PDFs and native CAD
• Standards and company rules embedded in a knowledge graph
• Detection of hidden dimensions, missing depths, and BOM issues
• CAD integrations with Inventor and SolidWorks for deeper checks
• Early drawing creation features and auto-tolerance suggestions
• Roadmap toward model-based definition and 3D-first checks
• Pricing is for setup plus subscription
• Bootstrapped traction, government support
• Long-term vision of CAD-less, decision-first engineering with fast simulation

Transcript

Host Intro And Guest Setup

Roopinder 0:19

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. My guest today is Uzair Sayid. Uzair the founder and CEO of NextCad AI, which has a very specialized but useful task of checking engineering drawings.

Uzair 0:45

How are you? I'm very good then.

Roopinder 0:46

Welcome to the show.

Uzair 0:47

Thank you very much for inviting me, to be honest. Yeah, pretty excited for this.

Roopinder 0:51

Oh yeah. And thanks for staying around late. I understand you're in the UK, London, I believe.

Uzair 0:57

It's not late, it's just evening, five o'clock. Yeah, pretty fine.

Roopinder 0:59

Well, I'm probably keeping you from dinner. I used to live in England, London, actually, for three years earlier in my

Said’s Journey From Pakistan To UK

Roopinder 1:05

childhood.

Uzair 1:05

Oh, you remember the borough where you lived?

Roopinder 1:07

I lived in Ealing. Where are you?

Uzair 1:08

So I'm just outskirts of London. It's it's called Bedford.

Roopinder 1:11

It was not cool to be Indian or Pakistani in that time. It's in the 1960s.

Uzair 1:18

Before I was born.

Roopinder 1:18

Before you were born. You're lucky to come there, go there when you did. You moved there from Pakistan, correct?

Uzair 1:24

Yes. So I came in 2008 for my degree, my degree. And then yeah, like the place, obviously. And then stay there forever, I guess.

Roopinder 1:33

They have good Indian food, just like home. So why not?

Uzair 1:36

Yeah, pretty decent. And to be honest, the quality is gone really good now. If you live for like 10-15 years in a country, you'll kind of like the food anyway.

Roopinder 1:43

Back then the food wasn't so good, but it was fish and chips. They didn't have the Indians A. So thanks for telling me more about the product. You have a very interesting application. You check drawings. Yes. CAD AI checks drawings, correct?

Uzair 1:57

It is very correct.

Roopinder 1:58

How'd you how'd you arrive at this idea? Tell me what happened. How did you get into startups? How did you think this was a good idea? Which it is, but tell me in your work.

Uzair 2:06

Obviously, my background is in mechanical. I did my master's, which I came for from Pakistan to hear in CAD. So I learned everything about CAD. For a decade, I worked for different companies, big companies, Fortune 500 companies. And as an engineer, like obviously I want to work on like, you know, solving real problems, figuring out new material, figuring out why things fail and physics and all that. But I think the problem I encountered in all the companies was wasting time on like documentation or you know, boring Excel sheet. Sometimes it's pretty annoying. Like you are even copying bomb from SolidWorks to Excel and then from Excel to ERP. This is not what we studied for. The frustration was always there. And in 2024, I started playing with AI and always wanted to get into startups. I thought, you know what, this is the right time where it's a level playing field for everybody. Nobody's 10 years ahead. So let's just tinker with. So I started playing with little automation, what can be achieved with AI. Now, obviously, back then the accuracy was not there, but it gave me the confidence that there are a couple of workflows, even in my past company experience, that we can definitely automate with AI, right? And then I list down all the workflows in my personal life. Like, look, this is pretty annoying. I was spending that much time here, that much time here, because I gone through

Frustrations With Engineering Busywork

Uzair 3:15

from a very graduate engineer to senior engineer. So I knew the whole process, even project managing, $40 billion projects. So I list down all the workflows, and I'm like, you know what? Drafting. As an engineer that loves 3D modeling, but I hate doing the 2D drawings, right? Same thing. Nobody got to automate it, right? So there were companies started doing the automation. I thought this always plays for a good player as well. So I started with automation of drawing, 2D drawings. Then we had some design partners, really good companies. And then once I got involved and looking from outside, previously I was thinking, like, okay, I am an engineer, what do I need? But now that I'm building a company, I'm looking from outside to these companies. I'm like, okay, every week we do meeting, we figure out okay, the bigger problem isn't the drawing checking. They waste a lot of time, even to find a tolerance from a data sheet from a third-party supplier. Engineer will spend half an hour, one hour figuring out spec to use, right? And then putting that tolerance in there. And if he's a junior engineer, he won't have that confidence to put the right tolerance. That I saw firsthand from outside, and I thought, you know what, this is the area that I want to first automate. So that's how a drawing checker was born.

Roopinder 4:17

Okay, got it. Interesting. So good idea. Now you have a good idea, but now you have to get some funding, you have to start a company, right? So you did that was a new thing for you. You didn't do that.

Uzair 4:26

That was pretty new. So the first thing that I I didn't know anything about the startup world. So this is me as an engineer quitting my job, and I'm like in 2025, I'll go all in. So in January 2025, I went to San Francisco.

Roopinder 4:38

Are you married?

Uzair 4:38

I am married, I have three kids. Your wife's saying, Oh, I got I still try to get her on board. She's on board, but it's like every time I know.

Roopinder 4:44

I don't know your wife, of course, but I would think that she would say, Wait, I married you because you're an engineer, and then you just put your job and you're gonna do what?

Uzair 4:52

Exactly. And that's obviously my from Pakistan, like very conservative mindset. We got stable income, we got a house, we got a nice car. What else do you need in life? I'm like, no, I need a bigger purpose. I can't just because I can see other people doing, and I can't really bottle it up in myself. Like, look, why not me? I tell to my wife, right? It's two years saving. Let me just do whatever I want to do in three years.

Roopinder 5:14

That's how I got her on board. I see. Now it's only been one year since you started, right? Less than a year. You just started earlier. Okay.

Uzair 5:20

It's I we actually I started so I went to San Francisco in the end of Jan

Origin Of NextCad AI

Uzair 5:25

for an event.

Roopinder 5:26

Okay. Yeah, yeah.

Uzair 5:27

So there is an event called Aleph. It's quite famous in the Muslim world. I thought, you know what, let's just use it as an excuse. Go out there, see the stuff, right? So, what's our startup? So went there, did that, and then luckily I get into Founders Inc., they have a two-week sprint going on. So I just said, you know, guys, please let me in. Uh, that's another story. Like, I just messaged them so many times that they said, Okay, yeah, man, come on. We don't know anything about you, but yeah. So that was the start of the startup. And then there, when I looked around, all these young people doing so like really good stuff, and I thought, you know what, I need to do it full time, can't just be half here, half there. So when I came back home, I told my wife all that story. Like, look, let's just do let me just do this for two years.

Roopinder 6:06

Okay, but you still got a little time left. This is good, and hopefully, you still have some money. Did you get some funding and investment then?

Uzair 6:12

From I was offered investment, but at that time I was because this was so new for me, and I was like, man, I don't want to take anybody's money. And I am like, because coming from an engineer's background, we don't want to take risks on especially on somebody else's money. So that was my mindset. I was naive. I wanted to have confidence in my product. We've done pretty well. Haven't raised the money yet, but we got funding from UK government which kind of supported us, and then yeah, and now the product we launched product in mid-October, and yeah, it's going pretty well. It's launched, yeah, yeah. It's launched mid-October, and we're already, I don't want to say, but we are already like default alive in a sense because we made enough sales, money in the bank, and that puts my wife to rest as well.

Roopinder 6:47

Hopefully, in one year, IOG, whatever you want in two years. So she's a little bit relieved, I imagine. Okay, good. Now it said you had party checked 2,500 drawings. Is that still a good number? Is it going up every day?

Uzair 6:58

It's going up every day. I think I put a tweet as well. So last week we did thousand eighty-two drawing checks. So it's definitely 50 per engineer. We made a conscious decision as well. We don't give out the software for free because the mistake that I did previously was we gave the software for free. People use it, didn't like the the output because obviously AI was new at that time, and they just left it there. This time we learned from our mistakes and said, until you then you use it because then they use and then they give us feedback, right? They give us a that's good. They say not working.

Roopinder 7:27

In this country, they say they have some skin in the game. Like I said, it's very niche, but it's very useful. And how do you I have to ask a general question first? It's almost a philosophical question. Did you think drawings were eventually going to go away and then you're fixing a problem that may not longer exist?

Uzair 7:44

Yeah, I mean, I thought about it, and definitely I think the drawing in one way or another, they're gonna stay there for at least the next 10 years.

Roopinder 7:51

Okay.

Uzair 7:51

Because even if you're not drafting the quality checking, the drawing is still considered as a sacred kind of contact between the manufacturer or a designer, right? I think it's not going away at least in the next five to ten years. But I get your point. The reason I'm doing this is so we can eliminate the drawings, right? If they go away, that's even better. The next thing we want to automate is the drawing creation bit. The drawing is created by AI, checked by AI, and I just look at it and approve the drawing and send it to the manufacturer. From the checking perspective, we definitely need to check the 3D modeling as well. If we can catch errors at the start point, that's even better, right? So we put model-based dimensioning. That's coming. If I survey 50 companies around me, or they're even using the AI checker, they're not using MBD, right? So it's a pretty new concept. It takes a lot of time to kind of dimension on the model itself than to create the drawing, right? So kind of negates,

Funding, Family Risks, And Early Traction

Uzair 8:40

but I hope we move this forward to the 3D model so we don't really need the 2D drawings. Or everything is automated. Creating the 2D drawing, automation, 2D drawing, checking automation. You're just making good models. Or even you don't need to do modeling. That is, even though I enjoy modeling it, I think that's still a waste of an engineer's time, right? You should be just making decisions, right? Apply this material, let's see what the simulation results are. Can we make this thicker? All these design decisions. That's what I'm like interested in.

Roopinder 9:02

Okay, so after more successful you are, the more likely that you won't need a drawing checker because the drawings will go away. You have a it sounds like you have a plan. What you're gonna do next, like you could check the model, check for compliance, do other things.

Uzair 9:15

Well, I have all these workflows that I want to kind of from an engineer's life, and pretty much like 3D modeling as an engineer, you want to make decisions, top-level decision and critical decision that will actually make the product better. Why we choose this material instead of this material? That thickness versus that thickness, you know.

Roopinder 9:30

How does it work? I submit a drawing. How do you take care of people's concern that their drawing is gonna leave the building? Is your service secure?

Uzair 9:38

We made local software which runs like their own CAD software and integrates with their SolidWorks. Well, we have two integrations at the moment, which is Inventor. Even giving them that kind of environment which is familiar to them, not going on the cloud, gives them that confidence that look, okay, nothing is going, or we're not taking any information away from them. And how we protect them is obviously we strip down any proprietary information or this drawing is attached to this company. We kind of strip that out. And also, yeah, I mean, we are not sending all the information anywhere. We are trying to solve this with traditional computer vision and algorithms. Eventually, the balance will flip, but until the AI gets really good, we're trying to solve it with the normal computer vision, you know, the segmentation OCR, and all those good techniques. So I would say 20% AI, maybe.

Roopinder 10:23

From your experience, it looks like you have a varied experience in hardcore engineering, but you also had to learn a lot of programming, a lot of coding. Did you taught yourself how to use what are you using, Python or C?

Uzair 10:34

As a kid, to be honest, I was really interested in programming, but I went to mechanical just because of my father, you know how things are back home. You gotta study doctor engineer, you know, software was not considered at that time. So I was always interested in software, and even in my engineering degree, in my bachelor's, I was taking software courses. I was doing assignments for my seniors in the software. I'm like, come on, guys, give me your assignment. I just love to do it. So that's my background. Officially, I am a mechanical engineer, but I used to do so many assignments for software that I was really good at it.

Roopinder 11:03

So you're hard, you're a developer, masquerading as an engineer. Interesting. Interesting. I never knew that's gonna come to you after 15 years or something. There was a time where they said software is gonna eat everybody's lunch. Mechanical engineers, who needs them, right? They just work on little nuts and bolts and gears. That's all gonna be replaced by software programs. How it works, I'm very curious. For 30 years, some years ago, 40 years ago, I started as an engineering intern, and all I did was make drawings. We had a person who would check all my drawings, and I used to live in fear of that person.

Uzair 11:34

Oh, yeah. When I joined, I was so scared as well. Like the guy who's the checker, he looks at my drawing and he's gonna bully me. I was like, no, I'm so scared. Right. Let's do uh thing. Now that you said, look, as an engineer, it takes us a lot of time to figure out what's wrong with this drawing on the surface.

Roopinder 11:49

It really likes it.

Uzair 11:50

Got all section views, got detailed views, all holes are properly lined up and everything. Now, what I'm gonna do is run it through my check. So this is a drawing checker which basically obviously lives in your desktop and you can upload PDF. Let me upload this. All you do is just drag and drop the PDF in this file. Okay. Let me just move this here, and then it runs the process. Let me just move us around as well. There you go. So it came up with center bore with the counter bore. So the depth is not manufacturable because we got the diameter of the bore, but we haven't got the depth, how deep we need to go for this diameter, right? So it picked up that without the depth of the bore, it will come back from manufacturing or production and it's emails back and forth what's missing in this thing. The other thing, and then it's also sets up the priority. The priority should be high, which medium, fair enough. Okay, and then these two are low to manufacture this part. We need the depth

Why Drawings Still Matter

Uzair 12:40

of how deep we need to go for this diameter. And the depth is not obviously the depth is only to that protrusion. Yes, exactly. They don't know how deep they need to go. Okay. So that's what it picks up. And also, like if it gives good feedback, you can train it, right? So you can thumbs up, and then also if it messes up, you do give thumb down. Well, we have like a knowledge graph, and you can just give it feedback. Like, look, you messed up, we don't need the depth, the depth is on this section. So the next time it runs, it knows like might need to look at other things as well. Majority of the things what's happening in companies are like so. This company, let's say this is a valve manufacturer, they're making hundreds of valves every day, right? Majority of the drawings are more or less the same. So that's another really good use case for this AI that it checks. You respond to it in natural language, you can give it feedback if it messes up, you can give it feedback in natural language. All that feedback gets collected, and we give company access to modify their knowledge graph as well. Because what you don't want is junior engineers actually uploading different stuff, which is then it becomes garbage in, garbage out kind of situation. So we gave like access to the to the admins so they can manage all these things, and also the rule set. So, what is it looking for? Like some companies will look for different stuff. So I can show you actually. Let me just see if I can show you.

Roopinder 13:46

So it's it's learning from my drawings and learning from your drawings, it's getting smarter and smarter. What is it coming in with initially? Have you a lot of drawings?

Uzair 13:54

Yeah, so when we started, we had like 10,000 drawings that we could trade a model on. It's not common database, and this this is another thing, like looking backwards. So because I spent 10-15 years in engineering, I was collecting all this like random drawings, or you know, I had in my folder for this company, blah blah blah. So I had like loads of drawings done anyway.

Roopinder 14:12

I I never knew that I'll use these drawings, but you know, yeah, they came in, they came in handy.

Uzair 14:17

They came in pretty handy, right? Yeah, and at the back end, right? So you can set up all these rules for whatever you want. So let's say if you are checking the machining part, so this is the part that I ran, and all these were turned on. It will check for tolerances, GD&T, it will see if the fits and limits are verified or not, thread specification. You can look for cheat metal, various stuff. Like in assembly, what do you want to do? Do you want to comply with all these? These are pretty standard, like industry standards. For example, in the UK, we use BS turned on. Similar for US customers, ASME Y14.5. Do you want to check your drawing against that? It depends on the industry. That's why we made it so customized. So it works for basically everybody. And also, what we have as well, so you can upload your own internal company document and then it collects all the rules from your own company document, and then it applies when it checks the drawing, it knows that these are your company rules. I'm going to check against these first, and then I will go to our next CAD, you know, model.

Roopinder 15:09

Okay, okay. So it's I'm impressed I can recognize things, features like holes.

Uzair 15:14

Now, is it also able to check one view against another? It checks against your views, it checks all the dimensions, it can read all the dimensions, it can read your bill of materials, and it can also read if a flag is like dangling. If you put a bill of material item, but you haven't ballooned the in the assembly,

Vision: AI-Created And AI-Checked Drawings

Uzair 15:30

right? So this is pretty common mistake that people do, or the balloon is like hanging in the air, so we don't know which it's referring to in the assembly. So all these little things because the good thing is because of my experience, I knew all these kind of mistakes or general mistakes that people do. And when I show, I mean it's very easy for me to kind of sell the product as well because when I showed or when I talked to the engineers, especially if they are on the in the on the car, they they thought they know, right? Oh yeah, yeah, we know we know this, right? And we want this. So it's it's an easy sell for me then.

Roopinder 15:57

Right. I understand that it seems to be doing a good job of checking annotation, but does it also check geometry? If I was missing a dashed line that showed a hidden edge, does it notice that you're missing a hidden line here?

Uzair 16:10

For PDF, I don't think it does, but that's why we have the CAD integration. This checker runs with CAD as well. I say I have this drawing. Let me just bring that and the checker. So now it's connected to CAD and it says inventor connected on the bottom. So we can read a lot more data. And if I go to the drawing checker, now it's checking the drawing against this. So it's not checking that it's automatically detected that there is open. I will check against inventor file, whatever is open. So now you're checking. Yeah, we we can read if somebody hand dimension if they hidden a feature or if they hidden something. So look at this hidden dimension value on sheet, right? So it's picking up there's a hidden dimension on the sheet, right? So if I were if you were a checker, you won't know that there's a hidden dimension. But you know, this picks up and it points me to the dimension that's been hidden. So look at this hide value, they hand typed it, right? All these little things we can read with PDF, it works really good. That does like 80% of the job. But if you want really good output, then yeah, we have CAD integration for Inventor and SolidWorks, and it can reach a lot more metadata. And then the idea is obviously we go to 3D model as well, so it can give you if you're let's say for plastic injection molding, your draft angle is not right, it's gonna have shrinkage.

Roopinder 17:13

That's pretty smart that you can integrate with inventor and SolidWorks too, correct?

Uzair 17:18

Exactly, yes. Inventor and SolidWorks we can do.

Roopinder 17:20

Hopefully, I'm thinking that on your wall somewhere you have a like a you know, like a temperature chart, only none it's not temperature you're detecting. You're saying how much, what percentage of the drawing you can check. Are you up to like 90%? You can check out the channel.

Uzair 17:33

Oh, second, like eval's perspective. Yeah, I quite could. I think if I were to put a drawings, it does pick up 80% of the mistakes. And obviously, with design intent, it's not gonna pick up the design intent, like why you made the shaft 10 millimeter, right? You could have made this unless it has the context of the whole assembly. We are obviously working towards that as well, that you can read the whole assembly in the whole stack and then make design decision as well or give feedback on the design decision. So we started like to be honest, this checker we started working on like four months ago. So we made quite good progress. The roadmap is quite big. I want to kind of make it so good that this becomes the industry default.

Roopinder 18:07

Yeah, I imagine eventually you want to not just suggest the changes but actually make the changes inside the CAD model, right?

Uzair 18:14

Yes, true. So we have a feature where it fixes the dimensions for you. So it taller suggests the tolerance and it fixes for you. It puts datum as well for you in the but it needs to be connected to the CAD software to do all that because we're calling a bunch of APIs to do that, right? So we still we we already have that functionality and we're kind of testing it to be honest. It's like we haven't gave it to the customer because then they will I don't want to kind of them this yet. I want to show you the other thing that we are working on as well. So let's say releasing it, we don't need to create the drawing. This is your drawing check, drawing creator. All I have to do is just say to it or just click a button, which basically says to the AI to create a drawing for me. And then it looks for and this is still in work in progress, so don't take it like you know, it's a finish for like it's but what I'm trying to show you is I look this the roadmap we are following is it's like yes, we want to eliminate the drawings. Then we want to eliminate even the 3D modeling, so engineers are like making high-level decisions.

Roopinder 19:06

Hey, you might be interested in this, but I just read about MIT taking videos, they call it VideoCAD, and they're taking videos of drawing creation. They're using art, they have 41,000 videos that can be

Security, On-Prem, And Minimal AI Use

Roopinder 19:20

used to train your AI on.

Uzair 19:22

Okay, oh wow, okay. Yeah, definitely interested. If we can get hold of those videos, that's pretty awesome. I could look at this drawing, right? It knew where to put the section, it also put your GD & T, your tolerances. So, even to find this tolerance, I need to go and find because it's a circlet, a functioning part. I need to go find in the supplier data set what's the tolerance, it automatically put the tolerance for me. And then again, it kind of knew that it's a shaft, something's gonna go in it. So let's put 87 on the tolerance, pretty standard. Again, you can change all these. It still messes it up, like it mentioned it twice. So you can make one reference or you're making some decisions on your end. But it to start off, it does pretty good job. Yeah, I mean, as an engineer, I would look, yeah, man. This drawing looks complete, everything is properly dimensioned, properly spaced, GD&T there. It missed the day for some reason, but all the tolerances are there.

Roopinder 20:06

I see, I see it's dimension to a part of the it's the on the on the end view, it's dimension to the outside, the diameter where there wouldn't be any material. It's over dimensioned, it's as dimension from the center and it's dimension from the outside where you couldn't measure it.

Uzair 20:21

Yeah, you you're right. So that's why we need to make so this is where like obviously engineer needs to kind of come in and say, look, make this dimension from the end, and then make this. If you want to leave it, leave it for reference, right? You know, but they then they're gonna take this dimension as the the driving dimension.

Roopinder 20:35

Yeah, well, I would have loved to have this. Where were you were you when I was growing up? I would love to have this in like last 10 years of my life as well. Actually, the checker, because as a junior engineer, you would love the checker, right? So, because it's picking up all the mistakes, putting giving you suggestions on tolerances and all that. So it does it. Are you a are the third-party program for Inventor and SolidWorks?

Uzair 20:56

And you they're we are standalone, okay, and we obviously connect with these CAD software. We plan to do it for on shape as well. This is one of the I think it's quite good. When I used to travel a lot with manufacturing companies, people were not used to or even open to OnShape because they didn't even consider it serious enough. So if it's NX, Inventor SolidWorks, but now I can see the shift, like people are using onShape. So the right player is emerging as well.

Roopinder 21:21

Now, as a user, let's say I'm using SolidWorks, do I just send you the file, the drawing file, or do I send you the model and then your service will come back with corrections? Is that how it works?

Uzair 21:31

That's how it works. So when you open a drawing in your SolidWorks and you have our software open, it will automatically detect that SolidWorks is open and a drawing is open. Then if you press run checker, it will check against whatever is open.

Roopinder 21:44

I see. So it can interface very neatly, it seems.

Uzair 21:47

It's seamless interface, yes. You won't even and even seamless onboarding as well, you put your configuration key pretty smooth.

Roopinder 21:53

I don't even have to exit my exit my part.

Uzair 21:55

My you don't have to exit, you don't have to exit your part. And once you run the checker, you can go and do other things as well in your cat software. You don't have to wait for the checker to finish its check. And you can run multiple times. So open one drawing, run the check, batch it, run the check.

Roopinder 22:08

How's your pricing model work? You charge per attempt, or is it all you can eat?

Uzair 22:12

Uh it's all you can eat, but we charge set of fees because obviously this is not for like very small companies. So it needs to be a theme of five and above. But we charge a set of fees where we understand their checks, what they want to check against on the back, and we prepare the software for them so they can extract the maximum value out of it. So we

Under The Hood: Rules, Standards, And Knowledge Graphs

Uzair 22:30

charge a set of fees, and then after that, it's subscription.

Roopinder 22:32

Got it, got it. And it's all it's all done with AI. You don't have a bunch of your old buddies from Pakistan. Oh, yeah, no. Tell me how it would work with AI because you've trained it and it sort of understands procedures and and standards, ISO, ASME, NC. So it's got a it's got a knowledge of that and it can apply. But do you also have rules-based checking? If this, then that is that both of them. Yeah, definitely.

Uzair 22:57

Our first priority is we check as many checks we can do using CAD APIs. So again, it costs us less and then giving everything the time, the response time is very less. So our first priority is checking against these single, for example, if I put this dimension across here, then that's the violation because one dimension is going to the another view. This we can do with API calls, right? Bunch of algorithms, clever algorithm to detect all that. You don't need AI for that. So all that we can extract from CAD APIs. And like I said, AI we are using like maybe 20 to 30 percent just for reasoning on some parts that obviously we can't do without with the CAD APIs. On the PDF side, there's no more calls from APIs on CAD software. So we have our own CRs algorithm, segmentation algorithm that we are extracting the same information or trying to extract the same information.

Roopinder 23:41

I see. Oh we'll work with my models. We already discussed this. And work with them with my models. I can also send PDFs to be checked. You would, of course, value the model itself because it's more checkable. You could more accurately check it and guarantee a better result. Is there a different price for each? Or you said it's all you can eat. Do you have two different prices for PDF checking or for model checking?

Uzair 24:03

No, I think we kept it really simple. Obviously, we just launched like two months ago. So we're still figuring out with the price, but we're trying to keep so that people are onboard with us and then they start using us, give us feedback so we can improve the product, right? So this is our main goal. But we don't want to distribute it for free because then, like you said, no skin in the game.

Roopinder 24:19

Everyone will say, Oh, it doesn't work. And yeah, they say free software is worth what you pay for it. You don't want to price it so high that the engineer has to ask his boss for money, they can try it themselves if they want to.

Uzair 24:31

Yeah, true. Yeah, that's one of the things that we're kind of figuring out how to give access to individual engineers. At the moment, the way we set up is going through the boss and then buying the software and distributing it among their engineers because of the setup fee, which is obviously a bit pricier. But we are still competitive with our competitors on pricing again because we don't want a price too high that we're not in the race, and we don't want a price too low that they think oh, this doesn't work, you are too cheap.

Roopinder 24:55

Do you want to say on the air what your prices are? Do you want people to check your site for that since it might be changing?

Uzair 25:01

Yeah, I want to not mention the price because we are still figuring out the price, and when you book a demo, we understand your problem and then we can price it up for you.

Roopinder 25:08

That makes sense. Tell me how you're getting the word out. I think I saw you first on LinkedIn, but are you using any other methods?

Uzair 25:15

No, to be honest, it's more like LinkedIn and my personal brand and my past connections. So far, all the companies that are onboarded were like more like, you know, okay, yeah, we know that he does a good job. Let's see what he's up to. So we're not going heavy on the marketing yet, but maybe next year we'll we'll try to go heavy on the marketing side.

Roopinder 25:31

Yeah, next year. I I have to say, once you can make a great product, there's a lot of smart engineers that have good products, and I'm seeing quite a few of them right now. It'll come to a point where marketing will take over the driver for success. Once you have a great product, it's fairly bulletproof, and now you just need to get the word out. I always point to SolidWorks as a best example of a marketing. They had a great product, but they were competing with Iron CAD and Inventor. They blew everybody away because their marketing was so good. You could even say all the MCAD products were similar. One was even better than SolidWorks, but it didn't matter. The marketing made everything happen for them. But anyhow, that's down the road. I have to congratulate you on already making revenue, which most startups don't want to. Some would even tell you they don't want to make any revenue yet, they just want to grow the market. But I don't know if that's very sustainable.

Uzair 26:21

Yeah, I want to make revenue, want to make stable company, right? So I'm thinking like long-term as well. So I'm not constrained by any investor or anything. Um that's my philosophy. I mean, obviously, I I can disagree with other people.

Roopinder 26:32

So the company is all years now, 100% ownership. How many people? 100% ownership, four people, four people, uh, and all in London or distributed?

Uzair 26:39

No,

CAD Integrations And Geometry Checks

Uzair 26:40

distributed.

Roopinder 26:40

Yeah, okay.

Uzair 26:41

Can't afford London yet.

Roopinder 26:43

Okay, but you do the bulk of the development and programming. Okay.

Uzair 26:46

Yeah, I mean, I do, I got very good developer as well, which I was working for like last two, three years. Um so the core team is pretty solid. And yeah, like you said, marketing, the first I joke around this, but I think this is going to be true. The first official hire is going to be a video editor or a storyteller or something like that.

Roopinder 27:01

I think my quality of video and audio is not up to my standards in engineering, that's for sure. I like what I saw. I wish you the best success. I'll give you one thing I've noticed, or one thing I'm hoping for, is that there will be a program, AI, that'll help you just design no matter what software you're using. It's just like I want to make a shaft. If you're using SolidWorks, it'll put in the right commands for SolidWorks to make that shaft. It'll interact with you. It'll say, How big do you want to make it? How long diameter, right? But it'll then produce the commands that the software understands. I know a lot of software programs, but each time I have to look up the command to do it. I wish there was something that would just be program agnostic, right? That there'd just be natural language interface to whatever I'm using. It could be on shape, maybe it's inventor, maybe tomorrow it's SketchUp because I'm doing something in AEC. Oh my god, that's a lot of commands to store in my head. My head's not good at that. I want AI help. Just like you're helping people with drawings, like AI, you figure it out. You read everything, you remember everything, read all the damn documentation and tell me what commands are in there. And by the way, put those commands in. So that's I agree.

Uzair 28:14

I think I would like to take it a step further and we say, look, we don't even need the CAD software, the future. In the future, why would you need a CAD software?

Roopinder 28:21

You're right, you're right. Absolutely. Take it all the way to the end. Put it. We say under the hood in our country. You say under the bonnet, probably, right? Exactly. Put it under the hood. I just want to steer the car. I don't care how the engine does it. Give me that high-level control. I want to be an engineer, I want to be creative. I don't want to get lost in. Oh, look up what command this is an inventor to do this.

Uzair 28:42

Too much so that's the future I want to build. You don't need CAD, you don't need all these drawing creation, drafting, blah, blah, blah. You make high-level decisions. Why this material? Why that thickness? What's the physics behind it? If we change this, what happens, why things break. And if I'm happy, approve, send it to manufacturing or send it to robot.

Roopinder 28:58

By the way, if we're gonna do it for design, we should do it for simulation too.

Uzair 29:02

Simulation is a big piece. And I think pretty decent companies like nTop, they're working on it. We need speed of as an engineer, want quick simulation, not preparing the model, giving it to another guy who's expert in simulation, and then he takes three days. No way. I've because I'm an engineer, I'm gonna make the design decision, right? I won't answer quick on that thing.

Roopinder 29:20

Exactly. All right, good work. I expect great things. Let you go. You're not making any money talking to me. But it was great. Thanks for showing it to me. Thank you again for joining me. Thank you very much. Thank you. Thank you for listening to FoDES, the Future of Design and Engineering Software how. 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.