#471 – An Interview with Matt Berggren

Chris Gammell: This is The Amp Hour Podcast, released December 15th, 2019. Episode 471, an interview with Matt Berggren. Welcome to The Amp Hour. I'm Chris Gammell of Contextual Electronics.

Matt Berggren: And I'm Matt Berggren from Autodesk, run the Fusion 360 team. All right. Welcome, Matt. How are you doing? Good. Good. How are you? Good. We are former co-workers. We are. We are. It feels weird introducing myself that way. I feel like I should just say, I'm business from Matt.

Chris Gammell: Right. So we work together at Supply Frame doing some product stuff there. Yep. Yep. Yeah. But you very obviously have a lot of EDA experience. We'll talk a little bit about your background, but you are here to talk about... I need a little bit of actually coaching on what the new title is. Sure. Sure. So it's Fusion 360, but also you were part of the Eagle transition, right?

Matt Berggren: Yeah. So I was actually part of the Eagle acquisition. So I coordinated the Eagle acquisition for Autodesk. I came in not as a part of Eagle, but actually to coordinate the acquisition. And really the emphasis now is on electronics design in Fusion. So I would hesitate. We did borrow quite a bit from the Eagle technology that we acquired, but what we've done in the Fusion environment is new, novel. There's aspects of what looks like classical Eagle kind of in there, but that's more a familiarity thing. And there are still quite a bit more than that. So Fusion has this history of integrating lots and lots of technology in it. So an example would be something like HSM works, which is used for like CNC machining, milling, cutting, that sort of thing. And when HSM, and I shouldn't say HSM works, that's a SolidWorks implementation of it. But when HSM was integrated into Fusion, we dropped the HSM moniker. And that's in large part because, well, it's Fusion, right? If we were trying to hide the strategy of what we're doing with Fusion, we could probably have picked a better name, like Diffusion or something like that. You could just call it Smush. You could have Smush. Exactly, exactly. So it was interesting though, because when we did the acquisition, I remember on EEV blog, other than being kind of taken apart for, oh my God, it's going to subscription. We hate you for going to subscription, blah, blah, blah, blah, blah. And that was later. That wasn't right at the point of the acquisition. A good friend from a university out in Pennsylvania had shot me a message over at EEV blog and said, I think what they're doing is integrating electronics design into Fusion 360. And if they do that, that's interesting. Curiously enough, he just shot me a message over Twitter the other day. And he said, hey, it looks like you guys are creating electronics design in Fusion 360. Yeah, yeah, yeah.

Chris Gammell: And actually, so give us some timelines on this. So when was the actual acquisition? And then-

Matt Berggren: So, and I remember this distinctly for what's probably a tragic reason. So the timeline was, we did the actual acquisition, closed the acquisition the day after the Brexit vote, which meant that we had taken a bunch of USD, bought a bunch of pounds, and then the following day, of course, the pound sunk, right? So- Oh, I see.

Chris Gammell: I was wondering why that was relevant. But okay, that's what you're saying.

Matt Berggren: You know, nobody can predict those things, right? Those are macro forces that nobody can predict. But- What is the actual date? I don't know the date of Brexit. So that was June 2016. So I joined Autodesk in January. You can imagine it was a fairly steep ramp up to both land the acquisition, the same time try to do a lot of kind of portfolio rationalization. Because when I came in the door, we had Circuits.io, we had 123D, we had Tinkercad, all of those rolled up into me. So, you know, trying to find a place for each of those things that didn't just sever the ties with this, with each of their respective user bases was critical. And so then likewise, you know, doing the Eagle acquisition, getting that in-house. And then you can imagine, I've got a room full of people who are working on Circuits.io. How do I sort of, how do I break the news to that group that, hey, we went off and acquired this electronics design technology. But I think what we've landed, both with the combination of stuff that we did with Library.io and then the stuff that we've done with the Eagle acquisition and the integration with Fusion, kind of is the best of all of those worlds.

Chris Gammell: Yeah. And remind people what Circuits.io is.

Matt Berggren: Okay. So Circuits.io was electronics design in the browser. And, you know, wherever you kind of fall on either side of that, probably the best example, best sort of analogy for people, if you just want to kind of orient yourself, is somewhere between fritzing and upbritter. Because it does have a breadboard editor. It does have, at the time it had schematic capture and PCB layout, though pretty limited. But it was really the circuit simulator underneath that. That was the really exciting piece. So that's actually on Tinkercad today. So I also run the Tinkercad team. And shout out to my Tinkercad team. Tinkercad is mechanical design in the browser, designed for sort of, you know, what we kind of euphemistically call like K to gray, which is a lot of pushing and pulling pixels and add subtract, right? So I take a geometry, 3D geometry, whatever it is, a cylinder or a circle or a sphere, a cube. I put them together. I add and subtract from one another. And I can build these reasonably sophisticated models without the same complexity that you might find in like something like Fusion 360, right?

Chris Gammell: No generative design. No generative design. Well, it's interesting. No server clusters and stuff like that.

Matt Berggren: Exactly. Well, you know, it's actually interesting though, because it's a, you know, all of those kind of 3D modelers in the cloud are largely based on, you know, WebGL, 3JS, these sorts of things. Oh, sure. And the distinction between mesh modeling and surface solids is lost on most people until you kind of get into that space and you have to hack your way through something like, hey, is this an edge?

Chris Gammell: Yeah.

Matt Berggren: How do I mesh these things together? And everything is triangulated.

Chris Gammell: Smush. Smush.

Matt Berggren: Yeah. Yeah. Exactly. You know, how do I, you know, I've got a bunch of triangles.

Chris Gammell: You know, Matt, I will license you that name. I actually, when you were speaking before, I actually, I registered the domain.

Matt Berggren: Excellent. Excellent. Autodesk. Of course, it's, you know, it's smush.sk. I couldn't get the premium version, but it's smush.tv. That's right. Yeah. So. No, that's quite different. But the circuit simulator that's on TinkerCAD is actually really interesting because it does instruction set simulation of the ATmega. So if you think about that in terms of integrating that with Spice, you have the analog simulation capability of Spice. But then if you have the instruction timing for the individual instructions, it actually kind of doesn't matter at that level what code you write on top of it. Whether you're running an RTOS or you're, you know, you're using Arduino drivers and services, like doesn't really matter because you've captured the instruction timing. So, you know, rising edge, falling edge timing, those sorts of things. And then you've got the individual instructions themselves and their sort of execution. When you put that in line with Spice simulation, it creates a whole bunch of really interesting kind of things that you can do with it. So that exists on the TinkerCAD front. You can go there, write code actually for a virtualized Arduino and, you know, other ATmega variants. There's even an ESP on the site, which is kind of interesting because you can have a device running in the browser and write code for the ESP8266. And that virtualized device can actually talk over Wi-Fi to an actual ESP8266. So it's, yeah, it's pretty interesting. You know, it's pretty interesting kind of a conceptual model, but you can imagine if you're a teacher with a classroom of 30 kids. Right. Handing out a bunch of Arduinos and jumper wire and all of these other things, you hope to push that as far down the pipe as possible. Right. Because just putting it away at the end of the class period is going to take you the last 20 minutes. Yeah.

Chris Gammell: It's a material nightmare more than, you know, the actual content. Yeah. I do remember the box of even just dead 555s and, you know, 7-4 series logic. Oh, God. Or even worse, the quote unquote working box when, you know, someone has zapped three out of every 10 chips in the box, you know. Yeah, exactly.

Matt Berggren: Exactly. So, you know, kids that are rubbing balloons against their heads and then stepping into electronics class, right?

Chris Gammell: Where'd you get that balloon? Where did it even come from?

Matt Berggren: Why did you bring it in here for God's sake?

Chris Gammell: We're using chips from the 70s, man. There's no protection here.

Matt Berggren: Oh, man. So, yeah. So, you know, then the, you know, that capability is something that I'm actually really excited about. Yeah. That sounds good.

Chris Gammell: And that's still under active development too. Yeah.

Matt Berggren: So that's still under active development that exists on Tinkercad. So we introduced something, let's say, a year ago, a little over a year ago called Tinkercad circuits. And, you know, the eventual goal being that at some point a kid wants to put LEDs in the eyes of a dinosaur that they design and run wire through three-dimensional space, we should be able to draw sort of conduit paths, for lack of a better term, for the wire to run through the body of the dinosaur. And then they can throw it onto a 3D printer, run the LEDs into it. And we have a few examples of that actually up on the website, which are things that are sort of plug and play. You can stick, you know, this unit into that body, do the subtract, and it will put an LED inside of something.

Chris Gammell: Oh, yeah. Yeah. So like just like a standard five millimeter or something like that. Correct. Correct.

Matt Berggren: So standard five millimeter LED, but also with like the coin cell in it. So 3D printable, nice kind of unit with a switch on the bottom or something like that. So you can put a coin cell in it, five millimeter LED will shine off of it. And then you subtract the whole unit from the body of the three-dimensional object. Yeah. And that has been really interesting. The uptake on that is huge. And every time that we show it to teachers going to something like ISTE, which is a big educators conference, their response is, you know, oh my God, where have you been all my life? And how do I teach this?

Chris Gammell: Yeah.

Matt Berggren: But so that capability, you know, that capability from Circuits.io was really the differentiated feature that persists today. And I think my, you know, my ultimate goal would be to continue to extend that technology so that on one side, you would have what exists in Tinkercad as sort of bootstrapping the market itself, right? Because there's market forces at work here. I don't want to shy away from that and say that we're a commercial business because I think there's too many people that do that. Um, yeah, we don't need one more in me.

Chris Gammell: Um, it's interesting thing too, though, like you said with teachers, I mean, you have kids who are designing electronics into stuff that 3D printing, 3D printing is just such a part of education these days and like in making, and then, you know, then you're going to have kids that are growing up, you know, even four or five, six years later, they're just like, well, wait, I can't put electronics in my designs. I think that leads very naturally into, oh, well, maybe I will, you know, take a look at

Matt Berggren: maybe we'll introduce this other piece into the curriculum. Yeah. And so Tinkercad, it's interesting, you know, Tinkercad has this export to Fusion and, um, just right at the intersection, the interplay of how do you make a mesh modeler talk to a solid modeler? Yeah. Um, so, you know, if, if, if you've ever played around in any of the sort of, uh, mechanical modeling tools, when you bounce in there and then you go, you know, to some website somewhere, you grab, you know, STL or some other, you know, OBJ, something that's been meshed, you bring it into those tools and you're like, oh my gosh, I can't do anything with it. Right. Um, we, we address that in a somewhat unique way. Now, Fusion has a mesh to what we call B-REP conversion. So B-REP is boundary representation. That's the solid modeling kind of, uh, standard form. Yeah. Um, and then the mesh conversion on that is really much more sophisticated. Um, but on the Tinkercad side, we, you know, we had tried to build a segue from mesh modeling, which is, you know, the OBJ, 3JS kind of representation of data. Um, and it turns out that the easy way to do that is to actually not store the model, but store the recipe. So if somebody draws a body and it's a cube and we know that that cube is volumetrically, you know, three by three by three, whatever the units are unit, unit, uh, uh, irrespective of units, the, if you store the recipe for how they got there, I did a sketch, the sketch looked like this. I extruded that sketch by this. Right. Um, you can play back that recipe and you know, there's, it's not rocket science. It's actually parametric modeling, but it's parametric modeling expressed in a somewhat different way where what we're storing in terms of parametric data is actually stored sort of behind the scenes, uh, so that kids don't have to see a timeline and scrub forward and back changes to legacy stuff and then play it back out to see how that changes things totally has its place in the, in the professional world. But if you tried to introduce parametric modeling to kids too early, it would be a little bit crazy. Um, but we did a few other kind of interesting things here, you know, added scratch for, um, uh, for parametric modeling so that somebody can actually build a sort of procedural model and say, okay, do this, do this, do this, do this, and now make that. Now let me go back and change the variable values and see what else it does. So all that's kind of on the Tinkercad side, but there is a bridge from Tinkercad into fusion that allows somebody to start there. I think that, you know, the inevitability of taking the circuit simulation capability that we have both in Tinkercad and in, in, in Eagle today and extending that to have more capability, more interactivity. Um, it's, it's the natural sort of bleed into, uh, what I would kind of, you know, hopefully characterize as like circuit analysis 2.0. Right. Like I think we've got to get past the 741 op amp. Um, I mean, I think that's a great example, but I think there's, there's more interactivity and I, you know, looking at what, what, uh, what my son has done in, in the program, uh, at Santa Cruz where he is, uh, the, the software that they use, um, you know, it still looks old and feels old and it feels like, you know, you look at it connecting wires and digital logic diagrams and things like that. And you think, gosh, you know, like this could be so much better. Um, I, and I mean, you'll, you'll probably recall when I was teaching all the workshops in San Francisco, I did a lot of that, uh, kind of pro bono. We can talk about that, uh, at some point, but the, the, um, I remember distinctly one time saying to a crowd of people in a workshop.

Chris Gammell: These are Eagle workshops, we should say. Yeah.

Matt Berggren: Yeah. Yeah, exactly. So teaching people how to use, how to use Eagle for electronics design and very practical stuff. You know, how do you solder a BGA? Right. Well, it turns out it's actually not that hard. It's probably one of the easier components to solder if you know how to screen solder paste, because you've got so much surface tension, you've got a decent, you've got a decent stencil. You know, it's like set it and forget it, right? Shove it in the oven and everything's good. But the, um, you know, doing those workshops, I remember saying to a class of people like, gosh, if they just gave me Eagle for like six months, I would rock this thing. Um, and that's, you know, that's in large part because of the history, right? Like coming from Accel EDA and then later Protel and then Altium and yeah, I've got this long kind of sorted history in EDA. Um, it, it was really fortuitous that all this stuff kind of came, came to a head when it did. And, and, you know, I got my comeuppance, I guess.

Chris Gammell: Yeah. Let's go back actually. Let's go back to where you got started first and then work our way through EDA. Um, so how did you get started in electronics in the first place? I mean, you, you're on the software side of electronics now, but you were, you know, long time hardware person.

Matt Berggren: Yeah. Yeah, absolutely. So I got started in electronics, uh, like so many people, you know, we, we always quipped that there's only two ways that you get into electronics, right? Ham radio or electric guitar.

Chris Gammell: Um, and I think that, yeah, the common one, the common one here is taking things apart, but yeah, I think that's true in both of those hobbies as well. Yeah, exactly. Exactly.

Matt Berggren: And I think those, you know, both of those hobbies are things that can be meaningfully improved by taking something apart. And especially with, you know, with, with electric guitar, improvement is not always improvement. It's a little bit more subjective. If I want to, if I want to overdrive this thing, uh, you know, I can, I can absolutely do some horrible stuff to it. Uh, but no, so it was, you know, some combination of that. It was right at the intersection of that. So, you know, I grew up in a house with ham radio. Dad grew up around radio most of my life. Um, you know, lots of very interesting, manually strong dipole antennas across the house. Uh, you know, don't touch both ends of that, you know, that sort of thing. Um, and you know, even today I go back to my parents' house as a dipole stretch across my bedroom from corner to corner. Um, but, uh, the, um, the, the early days of that were really around the guys who, uh, who were, uh, essential kind of contributors to the early days of Zenith. If you remember that. So my dad, my uncle, a number of other folks were all from Zenith, right? So, uh, television maker. Yeah. Zenith, like the television maker. So at some point that was a very progressive company, right? Zenith, Curtis Mathis. I mean, remember those names that like, like you have to be really old to remember Curtis Mathis.

Chris Gammell: I'm sure some listeners out there are. Yeah. But I, I, Zenith is basic. Zenith to me is like, oh, that's like the thing I could have pulled out of the trash because it was something, you know, an old enough TV that someone was upgrading it. But at the time, yeah, it was a great TV.

Matt Berggren: Well, they were, well, and they were pioneers in things like the early remote control. Um, so, you know, original IR remote controls came out of their R and D lab, um, you know, wired remote controls, things like that. So they, you know, they had a lot of quality of life improvements that they were making to the TV competing against companies like RCA. Right. Yeah. So, um, so I grew up in a household with, you know, a, a dad that was an engineer, uh, uncles that were engineers surrounded by lots of ham radio equipment and kind of, kind of progressed through that. Eventually hit high school. Rebellion was guitar. You know, the, the jokes on me because, you know, playing guitar didn't mean that I didn't learn more about engineering. That's right. Yeah. I can smirk and sit on the recliner and watch me, uh, watch me toil away trying to get overdrive out of something.

Chris Gammell: Oh no, not rock and roll with all that loud electronics.

Matt Berggren: That's right. That's right. That's right. You know, where's a solder? Um, so, so, um, yeah, so I kind of progressed up through that and then, you know, eventually, uh, really just committed myself to the hardware side of things in a big way. Um, you know, came up through the dot com period, uh, worked at microwave communications incorporated. I saw that you, for whatever reason, were reading a microwave book. You're going to have to tell me about that. Oh yeah. But microwave communications incorporated, if people don't know, that was, uh, MCI. So, um.

Chris Gammell: Wait, MCI, not like the phone company, is it? Yeah.

Matt Berggren: Like the phone company. That's what it was. I didn't know that. Okay. I didn't know that. MCI was microwave communications incorporated. So the original company was bootstrapped building, uh, microwave transceivers that would communicate between, uh, Chicago and St. Louis. Got it. Okay. Don't ask why St. Louis, right? I don't know. Maybe the arches. I don't know. Yeah.

Chris Gammell: Uh, but it's really flat to get there, I guess.

Matt Berggren: You know, pretty much, pretty much a line of sight, right? Just kidding. Can you just lift the pole another two feet? I think we're almost there. You just got to get real tall poles in your phone.

Chris Gammell: Yeah.

Matt Berggren: Yeah. Yeah. So gravitational influence on the, uh, energy moving in that direction. Well, arc it along the contours. Yeah. Anyway. So it's, it's, uh, um, it's, it, that was kind of what launched me into the dot com period. And, you know, MCI was a quarterback world com, world com famously, uh, it was one of the big players in the early dot com period led by a guy who very controversially ended up, uh, you know, driving the company into bankruptcy, but he, he was the guy that broke up the bells. That was, uh, Bernie. Oh, yeah. Okay. And he, he was the one that kind of opened up all of the telco infrastructure that was largely government funded, but managed by at that time, I think AT&T, um, and, uh, had, um, kind of was one of the key players in the early dot com, uh, by, you know, simply by virtue of giving everybody equal access to the telecommunications networks. Um, so that you can't plug into this network, uh, without making it available to people commercially. So, um, or you can't build out this network or build on this network without making the aspects of that, that you contribute to it. Something, you know, I, I, I will very much be reluctant to say like an open source project because I don't think anybody was really excited about building other people's infrastructure. But if you think about it, a lot of that, a lot of that infrastructure was in fact, you know, uh, built by multiple companies all contributing to the same backend.

Chris Gammell: Right, right, right. And, and really, I mean, as the consumer greater kind of, kind of a greater good in terms of just built more, more resources to be used, even if people didn't really know what they were getting in, in that process.

Matt Berggren: Yeah, exactly. Exactly. So, you know, and that kind of landed me out in California, you know, uh, the, the dot com out in California famously imploded. And then it was, um, it was, you know, as you were kind of watching the world fall apart, uh, I had been using schematic PCB tools for a long time, uh, knew some guys, uh, that had a product called, uh, PKed through a company called XL EDA used to be Excel version 14 and 15 and blah, blah, blah, blah, blah. Um, and, uh, and took a job in EDA. And I, I, I got the job, actually, it's interesting. You know, I got the job as, as a, uh, uh, technical consultant. It was a weird role. They were doing this road show where they're going to go city to city and talk about high speed design and, uh, microwave circuits and RF circuits and things like that. And all these things that you could do in PKed and like, like many EDA companies, and this is, uh, uh, probably not well understood, like many EDA companies that people that are really customer facing, um, don't necessarily have a background. Uh, I mean, this is, you know, this is, it kind of contributes to the imposter syndrome in engineering, but, um, you know, it's that there's a very famous quip from Kevin Morris who writes FPGA journal. It says that, you know, engineers are in constant fear of being found out. Um, you know, you put somebody who has been supporting customers, building electronics for 20 years out on the road and you say, okay, now, you know, we want to do a high speed design workshop. You kind of need to find somebody who knows something about that stuff. So, um, I kind of came in under that and then XL EDA was, uh, very quickly acquired by, uh, by then ProTel. Okay.

Chris Gammell: Um, so you, you were out doing, doing the roadshows and doing like demos and customer facing stuff.

Matt Berggren: So a lot of customer facing stuff and it, you know, you develop through that process, a real sense of both customer empathy, but then kind of self-awareness from a software engineering standpoint. So when you look at a piece of software, uh, and you've used it before, but then you've talked to 5,000 other people that are using it every day, you suddenly get this breadth of understanding of kind of what's the intersection of the things that are the most important. Yeah. And it turns out that it's not 44 layer probe cards, right? I mean, I would love it to be, but it's not, you know, the vast majority of products don't ship more than 50,000 pieces. Uh, the vast majority of circuit boards don't exceed six layers. Um, the, uh, the numbers of boards, certainly these days, this is, you know, kind of tipping in different directions depending on, on the technology that's, that's sort of in fashion, but the majority of boards don't use blind and buried vias. Um, you know, it wasn't until kind of mid two thousands, late two thousands, I'm gonna say 2007, 2008, that BGAs were approved for flight.

Chris Gammell: Um, we're approved for flight. You mean like, uh, like aeronautical stuff?

Matt Berggren: Yeah. Yeah. So, and that was in large part because the inspection was so difficult. Um, and yes, we could do x-ray inspection, but you know, the, the amount of contact, uh, for the material, making sure that, you know, all the pads really land where they should and that there, you know, there's enough strength in the bond between the individual balls and things like that. That was, you know, that was a fairly late, uh, uh, thing to come on market.

Chris Gammell: And I, I, you know, we, um, well, it sounds like a trailing edge type of thing too, where there's like a lot of confidence, you know, not bleeding edge. It's not the, I forget what all the different names are, but it's like, no, but you're right. Aerospace and like safety critical things are going to be trailing edge because they want

Matt Berggren: later to prove points.

Chris Gammell: Right. Whereas an apple or, you know, consumer goods in general would be willing to trade the, the size and the speed and the cost for just reliability. Yeah.

Matt Berggren: Yeah. Yeah. Yeah. And it's, you know, it really comes down to yield. You hope that you don't get to market with something that fails most of the, you know, these days chips are so much easier to test.

Chris Gammell: Yeah.

Matt Berggren: To ensure the fact that you've got quality. And, and, you know, there's, there's the, the other side of that, which is that we've just become far more of a consumer culture. So, you know, if, if a customer has an expectation that every two years or three years they're going to replace their iPhone, um, how, how much do you care about the, you know, 10 year bond strength. Vendor agreements. And yeah, exactly. Yeah. You know, it's, it, all of that stuff kind of changes. I mean, obviously there are still industries like that, that have these very long lead times. Um, I remember working with, uh, with some guys on, uh, on the emails project over at, at general atomics and general atomics used to be a part of general dynamics, but the emails project, which is public, it's not, you know, it's not hidden away, uh, is the aircraft launch system. So that, uh, magnetic aircraft launch system. So electromagnetic aircraft launch system is what email stands for. And it was replacing the old steam slingshot.

Chris Gammell: Oh, it's not like carriers and stuff like that.

Matt Berggren: And carriers and things like that. You know, you want to put an F-16 or F-18 up in the air. Um, the slingshot, especially the steam based ones, uh, are incredibly harsh on the body of the aircraft, uh, as are the arresters, right? You come in and, you know, you're catching a plane that's coming in at, you know, however the impulse is high, right? Or the jerk, the jerk is high. Exactly. Yeah. Exactly. But, you know, if, if you think about it on a nuclear powered aircraft carrier, uh, electricity is not a problem. That's right. So, you know, if you can leverage the, the, the nuclear reactor that you have, you can do some pretty interesting things. So they did some really, really interesting and exciting stuff. But I was talking with the guy and, and, uh, and he said, you know, we're building a system that is going on a ship that has a 50 year lifespan.

Chris Gammell: Oh, yeah.

Matt Berggren: The last, uh, the last rollout of this system will be on the ship that we build 50 years from today. So now think about that in terms of like just microprocessor, you know?

Chris Gammell: Yeah. Right, right. And being able to buy it and sourcing. Yeah.

Matt Berggren: Well, like my source code has to run on a processor that's going to be around for a hundred years.

Chris Gammell: Yeah.

Matt Berggren: Um, okay.

Chris Gammell: And yeah, and there's likely, there's going to be upgrades and, and modifications, whatever, but still, if not, you got to design for it, which is bonkers.

Matt Berggren: Right. Right. So, you know, that's, that's completely the flip side of consumer electronics. And it's, you know, it's interesting, um, the, the kind of dynamics there, but it, you know, without getting too far off track. Right. So, so, you know, meeting with the cross section of customers, you kind of find the intersection of what makes the most sense. And I think that really led me to want to contribute directly to the software and the product strategy that was going to help build the software that our customers needed. Um, and, and that was, you know, that, uh, I don't shy away from the history that I had at Protel or at Altium. I think Protel just for everybody's benefit, Protel eventually became Altium. And that was really a branding exercise. That was because we had PCAD, uh, and we had Protel and to sell two products within the same company at a similar price point to user bases that loved both meant that we were competing with ourselves internally. Um, somewhere was, excuse me, somewhere was, was the intersection of both of those products. And it was really PCAD strength and PCB at the time and Protel strength and schematic, um, that eventually kind of became somewhat hybridized in Altium designer and, and Altium was the, was the rebranding of the company. So, um, And when did that happen? That was early 2000s? Uh, yeah, early 2000s. So 2000, 2001, something like that. Um, so, you know, that I, you know, I rode that way for a long time and then left and, and eventually joined, uh, supply frame. Um, so left started.

Chris Gammell: Oh, you skipped, come on. You skipped the, you skipped one step in there. You used to work with Dave.

Matt Berggren: Yeah, I did. I did. So I did work with Dave. I did work with Dave, um, in the same office.

Chris Gammell: I was trying to, I was trying to explain this to someone earlier that like, yeah, Matt's coming on the show and, and, but like the fact that like you work on, you know, Eagle now fusion, right. And the fact that you worked with me and Dave, it's like, you know, just like these intersecting paths. It's just like, it's crazy. Yeah. It is a weird kind of, it's like a small industry. That's what it comes down to.

Matt Berggren: Exactly. Exactly. You know, it's like when you go to analog aficionados, which if anybody has a benefit of going to, it's pretty cool. It's like the who's who of Silicon Valley analog ninjas. That's right.

Chris Gammell: And it's just coming up in, uh, early February.

Matt Berggren: So there, I think that, it's such an amazing event. Yeah. If you haven't, if, if you haven't gotten a chance to go every year, it's kind of the who's who of the people that invented every analog device that is, that just, gosh, you know, has moved the needle and moved the industry forward in huge ways. But yeah, it is, you know, you realize how small the industry is. Um, but yeah, so I did work with Dave. So I, so, and, and for whatever reason, I spent the majority of my adult life after electronics, uh, and I shouldn't say after electronics, that's kind of misleading. I built electronics through that entire period, um, either directly or indirectly working with customers, work, uh, working with customers on behalf of the company or working, working, working with customers outside of the company and just building stuff, uh, contracting to folks with just a whole variety of different things. But the, the, um, the, the, I, I spent the majority of my adult life actually outside of the country. So I, I moved to, uh, moved to Australia, spent, you know, better part of about five years in Australia, uh, where Altium's corporate headquarters used to be in Sydney. Yep. Um, and then eventually we relocated the corporate headquarters to, uh, Shanghai and that was in part fueled by a couple of different forces. Um, you know, there, there, there were a few different things at play there. One of which was, uh, the obvious, uh, Hey, here is this burgeoning electronics market. They're graduating 500,000 electrical engineers every year. Uh, we have incredible market penetration, but not a lot of revenue. So, so what that translates to is we have an intellectual property problem, right? People are leveraging our software, using our software, and we need to be there on the ground. If we want to have a credible, uh, opportunity to convert those customers from non-pick to pick.

Chris Gammell: And right. Exactly. Show value.

Matt Berggren: I mean, just engage with people and say, Hey, listen, you know, we know that you've been knocking off the software, but you're a really big company. And, you know, we kind of like you to sign the guest book. Um, um, but then there's, you know, the other side of that was, is that they were graduating, you know, half a million engineers a year. And there's this huge opportunity in that space kind of happened right at the time in which Intel was throwing around, you know, numbers of trillions of devices that were connected by a IOT and, and it, and it made a lot of sense. Right. And we are getting there too.

Chris Gammell: I think that was like the right, the right number, but the wrong timing, you know? Yeah. A consultant dropped like a, like a date conversion or something like that. Yeah. Yeah, exactly. Exactly.

Matt Berggren: And, and, and, and, you know, it was, I think it was the right number. I think it was, uh, it triggered unnatural acts to try to accelerate that, which I don't think were necessary. Yeah. I think we were going to get there anyway, because every television was going to be connected to the internet. Every, you know, every whiz bang thing around their house was, was eventually going to have internet connectivity. But the notion that, uh, connecting those things together, uh, was something that had to be exceedingly deliberate. You know, I, as much as I like my Amazon Echo, um, I, I, I don't know a lot of people that connect that to a hundred different devices around their house.

Chris Gammell: That's right.

Matt Berggren: So I think we still haven't gotten to the aggregate intelligence, right? We've gotten to the sort of, yeah, it's connected to the internet. Great. It can stream some audio.

Chris Gammell: Right. That's pretty much what mine, mine starts music for me and keeps kitchen timers. That's, that's about the extent of it. Mine too. Mine too. It does great at that though. I really like it, you know, like from that perspective, I don't like the listening thing, but like the fact that it's just there and easy. Yeah. It plays music for my dog all day long. I also like that. It chills him out. He's listening to some Christmas jazz right now. You know, great, great stuff.

Matt Berggren: Yeah. And I think, you know, it's, it, it was somewhere in there, there was the, there was just the mistaken impression that that somehow required us to all mobilize lots and lots of resources. And I think, you know, there were lots of examples of startups, especially that were built, funded on the backs of, oh my gosh, IoT is coming. We're going to build something, which is easy to add wifi to something. I don't think that that's altogether flawed. I mean, Espressif with the ESP8266 is pretty, pretty sweet. Um, and, and, you know, having lived long enough, um, when there was a period there where we went from serial ports to USB, uh, rabbit semiconductor, uh, which these days I think are the guys that do, uh, the, uh, a lot of Zigbee related stuff. Um, rabbit semiconductor built these USB to serial converters. Um, and they were plug and play modules. Now, you know, on the backs of that FTDI emerges, right. And you've got people who are building single chip USB to serial converters or serial to USB converters. And, uh, and all of a sudden an industry, like, you know, we were talking earlier, uh, off, uh, uh, outside of the conversation about McDonald's, all those early POS systems that were used in the restaurant, the point of sale systems, they were all serial ports. Yep. Exactly. Everything needs to be updated. And those required amplification and, you know, you got signal amplification because you're going to the back office. The back office might be, uh, you know, 20 meters away. So, you know, you've got a reasonable length of cable there. You've got signal amplification and attenuation issues. You've got all sorts of stuff. And people wanted to just plug into the back office computer.

Chris Gammell: The, the, the, the reference here is Matt used to work on some of these things as well in his Chicago days. I know a little bit about it. That's what he was saying. He said before. So I just wanted to clear that up. Yeah. So, yeah. So Matt was working on this stuff. And, and so, yeah, you're saying that the connectivity piece was based in wires and physical things. Wires and cables. Yeah.

Matt Berggren: You get somebody out that would drag cable. And these days, I think, you know, the, the, the translation of that, this sort of transition is toward wireless connectivity. And the expectation is wireless connectivity. Um, you know, today when you pull up to a drive-thru.

Chris Gammell: Well, I think there's still like a, uh, you know, a business need there too. Right. It's not like they were doing it to be frivolous. Right. McDonald's is a very good example of like, no, no, no. We have like thousands of locations. We need thousands of ways to quickly interact with customers. And now we can upgrade our systems and doesn't have to be wired anymore. It's like, okay, we still have that need. Now it's just new technology layered on top of that.

Matt Berggren: Right. Right. And I think it's, it's the notion of that being something that, uh, was super deliberate. We all have to go out and somehow raise money and build businesses around that. Um, you know, I, I, I did that. Right. So myself and the, uh, then CEO of Altium and I left Altium in 2012 and started a business in China really focused on IOT with the goal of connecting, uh, infrastructure together. Cause you know, here, here you're in this country that, you know, in Shanghai alone today, there's 30 million people, right? Most populous cities in the world. Right. 30 million people. Most states in the U S are less than that. Exactly. I'm pausing on that for a moment. That is the population of Australia.

Chris Gammell: Oh yeah.

Matt Berggren: There you go. So, um, yeah. Yeah. So, so you concentrate that in one city, you've got this country that's going through the largest mass urbanization in history. Meaning people are moving out of what were historically villages into cities looking for not just a better quality of life, but, uh, looking to level up in terms of skill capability, um, and, and, uh, and contribute to the economy in a real way. Um, and over there, I think that's a very real, from my experience over there, I'd spent better part of about four years there. Um, that's a very real, uh, undercurrent that's underappreciated by people. Is the interest in sort of mutual contribution to the economic growth and success of the country. Um, it's, it's, you know, very real and tangible, uh, kind of, uh, need for building out this technology, this functionality, these features that would, um, you know, connect together the infrastructure that's there. Um, and, and that was in part informed by some experience that I had doing, uh, some early R and D, uh, for McDonald's. Um, and that was a, you know, that too was kind of an interesting project. It was, you know, largely focused on the restaurant and how do you connect together the internals of a restaurant, um, in a more intelligent way? And I, uh, you know, hats off to the guys at McDonald's. They were pioneers in this long before. I mean, the, the 900 megahertz Zigbee spec wasn't dry yet. Yeah. And they were asking questions like, how do we stop guys dragging cable through the restaurant?

Chris Gammell: Yeah. Right.

Matt Berggren: Because to do that shuts down the restaurant. Right. How do we get all of our franchisees connected into a back office computer that connects to some backup, uh, some, you know, larger back office infrastructure that eventually backs up into our corporate headquarters. So we can observe that somebody didn't just add, uh, you know, five rolled tacos, rice and beans to the menu at McDonald's. Um, you know, these, you know, it's really about discipline for them and it's really about data.

Chris Gammell: I mean, like that's an early, early focus on data as a. Extremely. As a, uh, resource and a, you know, not a commodity, but like a valuable thing, you know?

Matt Berggren: Well, and, and they acknowledged, especially when they were trying to pull together, uh, McCafe. Oh yeah. Um, I think rightly so, um, that, uh, the ambient noise level in a restaurant at that time was just nuts. It was like going into a shopping mall. Um, and, and you really need to bring that down. If you want somebody to sit and have a cup of coffee, you need to bring that down. If you want to compete with the Starbucks of the world. And I think everybody at that time was trying to figure out, you know, what is this Starbucks thing? And, and, you know, what are the implications of, uh, this, how does this affect our business? What happens if they sell a sandwich? Right. So, um, you know, I think the, the stuff that we had, uh, endeavored to do in China was kind of the, the next generation of that thinking, at least for me, which was how do you connect infrastructure together? So, um, you know, if, if every light on the highway from Los Angeles to Las Vegas is lit all day or all night, um, is that really necessary if the rate of trucks are, you know, one every 60 seconds, um, and can we connect the lights together to follow the truck along that path? Can you look at, uh, then certain events and respond to them? A truck that is picked up by one sensor, but then immediately, you know, not picked up by another such that you can sort of predict that something catastrophic happened in the period between here and here. Um, because nothing slows down that fast. So, you know, it was those kinds of interesting problems. Uh, you know, I think I mentioned to you at one point, the, the, um, smoke detection stuff that I was very passionate about, which was, you know, can you connect an HVAC system, which has some fluid dynamic simulation behind it with, uh, with the smoke sensor, with, uh, temperature sensing, with proximity detection. And it turns out that at the intersection of that are some really, really difficult problems of, as the temperature goes up, you know, all of the, all of the electronics behave differently as smoke starts to appear. What is, uh, a person in proximity to the proximity detector versus, versus, you know, just smoke clouding the area. Um, and as any good hacker knows, right, the, you know, the dust spray that you use for your keyboard is an easy way to, to trip sensors that think that there's a human on the other side of it.

Chris Gammell: Yep.

Matt Berggren: Um, but those sorts of problems become really, really interesting in that space. And I, you know, I really wanted to solve a lot of those problems. I, you know, tragically, I don't feel like that chapter really ever closed for me. I feel like there's still an itch there that's got to be scratched at some point.

Chris Gammell: And I, I don't think, I don't think sensors and or connectivity are going away anytime soon, Matt. So I think you got some time. Yeah. So I got some time, right?

Matt Berggren: I got some time. Um, you know.

Chris Gammell: Well, it's interesting too, with the infrastructure idea too, like you think about deployment timelines, right? We talked about the 50 year timeline of a, you know, of a ship, but when you think about like, not only like getting things out in the world or, you know, to use your example of sensors on the roads, right? Yeah. But maybe, you know, Phoenix is different than Tucson is different than, you know, Flagstaff and, you know, like, and, and really what it takes is a bunch of churn and struggle. And then eventually some kind of general consensus of like, I guess we'll just all use this one thing, which is not, still has not been decided. And so, yeah, I think there's a lot of, a lot of space for that still. There's some, well, and I think definitely churn.

Matt Berggren: There's yeah, definitely, definitely. And I think, you know, we, we still haven't decided what to do with the aggregate intelligence for all of those things. Yeah. Right, right, right. Because nobody's agreed on what that means.

Chris Gammell: Usually Black Mirror, Black Mirror picks up on it first.

Matt Berggren: You know, we still got a lot of work to do on that front. I was at a, an IoT conference and just as a panelist and that, that asked us on the way out, Hey, you know, what, what do you think needs to change in the industry? And I said, I think we need to be exceedingly transparent in what data we're sending. And by that, what I mean is, is, you know, don't tell somebody that it's, you know, Oh, your data is encrypted and it's coming via open SSL or something like that, or SSL, you know, let's, let's be, let's be very, very different and tell people we're going to collect this information. It's going to be sent to these people. So there's a level of accountability so I can make a choice. And I don't think that that is unrealistic, but I think it's unrealistic, right? I think, I think the commercial opportunity is just too great that unfortunately you don't get, uh, you don't get a lot of, um, control over those things. Um, in large part because the data is too valuable. Um, so there's a lot of negative forces.

Chris Gammell: It's very pervasive. I think there's a lot of value in it, but I think that you're right. It's like, uh, it's, I mean, again, it's a resource and it's something that people should, should know what they're, what they're signing up for.

Matt Berggren: Yeah, absolutely. Absolutely. Absolutely. So anyway, you know, fast forward and I, you know, I, I, uh, relocated back from China to the U S and, and, uh, decided to, um, you know, at that time I'd been, uh, spending a good deal of time hanging out with my good friends from supply frame, uh, who had started conversations with long before that at Altium, uh, who, you know, has really focused very heavily around vertical search, uh, targeting the electronics vertical guys that were, you know, in part bootstrapped from the people that, that founded companies like Abnet and drove, you know, Marshall Electronics to, to eventually, uh, become one of the first online retailers of electronic parts. And, you know, it was just that the story there was so interesting because there's so much in the supply chain around electronics that's poorly understood and taught nowhere. And having gone through the process of building commercial products and, and, uh, also higher volume consumer products, it, you realize that there is an art form in just how you capture things like attribute information, um, such that the person downstream who's on the receiving end of that can make decisions without too much constraint. And what supply frame, I think, uh, opened my eyes to was just how important and essential a part of the product design and manufacturing experience that that was. And, um, and at the same time, supply frame is just an amazing company to work. I mean, you, you, you can, you can attest to this. It's a fun place to work. They're super interesting and fun people. Yeah. You got a lot of hackaday stuff too.

Chris Gammell: That's a lot of fun. Exactly. Exactly. You and I and Sophie and Alec and a bunch of other people helped, you know, that, that first, first super con back five years ago. So that was a lot of fun.

Matt Berggren: Yeah. I remember my planning days. That was, you know, that was, I mean, but it was such a good time, you know, such a good time. And it's, um, you know, I look back on that with just, just incredible fondness because it, it, when you, when you want to be at that intersection, I think of gathering data and intelligence and then leveraging that to do things like, uh, drive a better advertising experience, drive better access to content and resources, you know, data sheets, reference designs, those sorts of things drive better sort of pricing intelligence, uh, you know, exposing pricing from a bunch of different places and kind of aligning those things to one another. Um, you really have to do right by the, by the community. Um, otherwise you're at risk of being viewed as the company that gathers lots and lots of data and, uh, and, and you're totally opaque to the user. And I think, uh, supply frame really struck that right balance for me. Um, otherwise I don't think, you know, if it was like a Lexis, Lexis, nexus, I always trip up on that. Um, you know, I don't know that I, you know, that I would have had the, the enthusiasm for something like that, but I also really like just trying to understand the big search problem. And they certainly went to extraordinary lengths to build out the infrastructure to handle some of those problems.

Chris Gammell: So, and I think, well, I think the data, the data thing, like you said too, I think that that's a good lead into the, you know, Autodesk Eagle days, like, you know, a lot of the, uh, data that's being captured now for, you know, on, you know, more web connected tools like, like fusion is right. It's like, you get more, you get value out of that because now you're tracking, you're tracking each and every move and you're, you're seeing that you have that networked. And there's also connectivity between, you know, between different users as you want to do, uh, sharing tools and stuff like that.

Matt Berggren: Yeah. And I think there's, you know, it's, it's interesting, you know, we talk about cloud data as though it's somehow this foreign body. Um, and, and the truth is, is that people have been putting stuff in version control for quite a while. Right. I mean, since, since, uh, you know, the days of RCS, which is a revision control system. Right. Um, but, uh, but I, I, there's nary an engineer doesn't have a GitHub account these days. And certainly your software engineers all have something in Git somewhere. Yeah. Um, or it's in Perforce or it's in something else, but a lot of these are hosted solutions. A lot of that is the intellectual property that are in fact, the keys to the castle. I think what we do with that data and, and Autodesk, you know, it's interesting because it's, it's not really talked about outside of the company very much, but one of the things that, um, comes up in conversation periodically is, you know, well, what are we, what would we do with the data that we start to understand about customers? And the feedback is constantly, or the, you know, the, the conversation amongst the executives, and I can only say this as a recipient of that is we, we will only do those things that make sense to add value to that user.

Chris Gammell: Right.

Matt Berggren: So it's not about taking that data and in the aggregate somehow analyzing it and, and performing some set of things on top of that. Uh, it, the level of detail that we get is also, uh, not to the level in which it's actionable intelligence. And, uh, in, in many cases it's, um, it's, you know, does somebody use the software? How frequently do they use the software? What, what parts of the software would you use? So that is actually one thing that I, yeah, that's one thing that I learned from the supply

Chris Gammell: frame days is like just how spotty data is in general. Like not just, uh, you know, what's out there and what's available, but also like you and I might be both be researching for resistors and, uh, you know, we both want, you know, 49.9 K resistors, but there's, you know, so few people searching for similar things at the same time that like correlating is really tough. And so I think what you're saying here is the data that you get around me as a user might be valuable to me, but it probably isn't as valuable to compare me to Bob down the road to Jane, you know, a couple of miles away from that, you know, like it's more of a individualized experience.

Matt Berggren: Well, and I think even, yeah.

Chris Gammell: Yeah.

Matt Berggren: And I think even more explicitly, um, it is wrong for us to triangulate user information in a way that, uh, is leveraged, uh, to try to drive user behavior, uh, other than to drive user, uh, understanding and know-how and capability. So, uh, for example, like infusion today, um, there's a workspace panel within the software that, uh, when you open an editor knows that you opened, for example, like the cam workspace. So, Hey, this person cuts metal. Great.

Chris Gammell: Yeah.

Matt Berggren: Uh, you probably want to know how to use this. Um, that in and of itself is built on your interactions, but it's, it's targeting you, the individual, not targeting our, our user base in general. Right. Um, if, if there was something to be pulled out of that as like aggregate intelligence, then I think we have an obligation to our users to be transparent about that and say, Hey, do you want to explicitly opt in to receive something which is going to take your information, collect it with a bunch of other information to give you the toolbar configured the way that most users leverage the most capabilities of the software. Like that would be a, you know, kind of an obvious double click on that. Sure, sure.

Chris Gammell: You're saying like, uh, like a float to the top kind of thing of like, this is the most, the most popular feature kind of thing. Yeah. Right. That makes sense.

Matt Berggren: Exactly. And I think, you know, part of, uh, part of what I've tried to do in the Eagle UI, uh, classically, it was just all over the map. And since we've integrated electronics capability into Fusion, and it sort of look at the process as a workflow oriented process. So before, you know, you had a toolbar and you kind of group things, you know, kind of in lumps, right. Uh, you'd say like, okay, these are all the kind of line drawing related things. And these are all the, you know, this thing. Um, but it's really about, you know, how do I define a board outline? So, well, how do I set up my layers, define a board outline, place components, route them together, uh, and, uh, you know, pour my polygons, rip things up and rework them. Right. Like if you just sort of put things on that continuum, uh, then you can organize things in a way from left to right that are prioritized based on how most users would do things. And, and, and all of that is just coming from the experience that I've had and, you know, talking with, I don't know, God over the course of years, because I, I, you know, I was in a tech sales team. I was field apps guy. Like, you know, I spent, when you come up through a company, which is relatively small at the time, I don't know, $30 million, you know, growing that business over, over an extensive period of time. Um, you know, today I think their market cap at all team is what three, three and a half billion or something like that. So it's a, you know, it's substantial player now. Yeah. Yeah. Um, where before we were, you know, we were the pirates, right? Yeah. Yeah. We, we, we didn't have a right to play. Um, and, uh, and we had to do a lot of work to, to, to ensure that we did get the right to play.

Chris Gammell: So, you know, when you look at something like customers and a lot of listening to customers

Matt Berggren: going in there and, you know, getting the pricing just right. So that they could get it in with their approval. Yeah. Um, being sure that it was easy to install, easy to use, that it would run on just a desktop

Chris Gammell: operating system. Less than 5k goes on a credit card. And so it was priced at $49.99. Exactly.

Matt Berggren: Oh man, it was very deliberately priced at $49.99. And that was because we had that real sense of customer awareness. And, and there were lots of certain, lots of circumstances where people would say, Hey, you know, we, uh, just discovered that all of our engineers are using your product, even though we have these really expensive licenses of this or that. Um, uh, can, can you tell us exactly who you are?

Chris Gammell: Right.

Matt Berggren: And so, you know, you, you sort of win by being scrappy by that. Well, you know, almost invariably you win by being scrappy, but it's the, you know, it's the willingness to be a little bit humble, acknowledge your, your, uh, you maybe don't have the right to play today. Um, push the envelope in terms of technology, capability, functionality. And, and I, you know, my hat's off to the guys over there. I think they've done a really good job of that. Uh, even after, um, after my departure and the departure of, of a lot of the guys from, uh, from, from the very, very early days of the company, um, you know, continuing that, that push towards trying to build functionality and features that are relevant to the user. But you've got to do that quickly. You've got to do that in a very pragmatic way because you've got limited resources. And I think what, uh, what we're doing today is actually kind of a reflection of that, uh, that experience, at least for me.

Chris Gammell: Yeah. So let's talk about that actually. So, so you mentioned, you know, the Altium days with scrappy had limited resources at the beginning and now you're coming into Autodesk. You, you know, you have this new product and then I think you have definitely the resources, but then maybe I'm curious actually of, of, uh, how much goodwill was there? Like, so like when you did decide maybe you wanted to move, uh, you know, the tool grouping or similar things or, or change some UI, you know, what was, what was the feedback like? Cause it feels like Eagle had, has a very strong, uh, loyal user base that are maybe resistant to change.

Matt Berggren: Well, so, so I think if, if there's one, one reality that I came to grips with teaching workshops, teaching people how to design electronics, it was that, uh, people didn't use Eagle because of the ease of use in the UI until they got past a certain point.

Chris Gammell: Yeah. Yeah. Yeah. Sharp learning. And it was, I know this one well, man, I know this one well.

Matt Berggren: Well, and, and once you get beyond a certain threshold with Eagle, all of a sudden you realize what's good inside of it and why it was designed the way that it was. It took me a long time to figure out that, you know, that the, for example, the layer names, you had names like tea paste and, uh, uh, I play on tea cream, of course, right? It's tea pain. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. But you know, tea cream, bee cream, uh, and I'm, I'm, you know, it puzzled me for a long time. Why do I call this cream? It's solder paste. Come on. Stop calling it cream. Here's the reality. Software, uh, Eagle, Eagle is, uh, it's native compilation environment with sent OS. And if anybody knows anything from the next days, uh, you know, Unix, Linux, whatever, um, the two character, uh, shortcut in Linux, Unix, whatnot is the, the, the magic, right? So, uh, yeah. So do. Right. So super user do. Right. So the SU being super user, giving you that, that, that, that leverage. Yeah. So LSD, you name it. So, um, the, um, the layer names, because the command line in Eagle uses a starts with function, the layer names disambiguated so that if you typed something, you only had to type the first two characters of it and you would end up with the layer that you wanted. Uh-huh. Now that's magic for somebody that knows, but it is completely anathema to everything else. That's right. That's right. It's like, why is it called cream? This doesn't make any sense. God, this doesn't make any sense. And then you're like, oh, well, there's no other TC in the list of layers. Yeah. They picked something that would disambiguate. Yeah. And, uh, and that took me a really long time to understand. And some of those things I didn't find until I was in the source code. Oh, interesting. And, and, and, and, and until we really brought the, the cat's off guys in or what were, you know, uh, the cat's off guys into the organization and kind of really tapped them on the shoulder and said, you know, why this? Yeah. Um, so, you know, in, in all of these applications, they're fundamentally very, very similar. You know, you've got a command queue that handles commands from UI events. Uh, you've got a UI framework or UI toolkit that builds the front end of that thing. And then you've got some underlying state machine that processes that, and you've got a data model for, for how you store that, uh, both resident memory and on a file system. Right. So there, you know, if I have to take the sum total of all of these experiences and kind of narrow them down to, to, to the very basics that those are the very basics. So, you know, if you're taking your computer engineering course and somebody's having you do state machines for the first time, um, rest assured that is the rest of your life. So if you don't like it, I can bail now and go and start, start studying that, you know, finally take that literature class that you've always liked. Yeah. No, no, but, but, but, you know, being very serious about it though. Um, you know, that's, that's kind of the fundamental application architecture. And when you, when, when we looked at what the Eagle guys had in particular, the core of the application was very, very robust, very well-written, uh, extremely deliberate, very thoughtfully put together. Um, and, uh, and the user experience, um, was something that the vast majority of new users really struggled with. Now the legacy users, you have to imagine that in any software application.

Chris Gammell: Well, especially a long, a long living one. I think that's a key point too, right? Like these days apps live for what, six months, you know, like, oh, I couldn't install an app on like a tablet that was a couple of years old, you know, like a couple of days ago. Yeah, exactly. Yeah. You know, I mean, how many people are still playing Angry Birds, right? Yeah.

Matt Berggren: Or bejeweled or any of those things. Right. Um, now at one point that was, you know, it was all sorts of fervor. I mean, they made a movie about it for God's sake. Uh, so, so, you know, the, the, the reality is, is that, um, you do have a longstanding user base, but over time, uh, you lose those users. In the renewal process. Right. So, especially with desktop software that is non-subscription, and I'm not going to shy away from that, um, you lose those users in the process. So you end up with a pretty substantial user base that just doesn't renew anymore.

Chris Gammell: Right. Right. And that's what it's all. You're saying from a, uh, you know, like who you're then developing for is kind of that. Yeah, exactly.

Matt Berggren: So you're always developing for, yeah. So you're always developing for a new base of users. Yeah. Um, you know, you, uh, over any period of time could, you know, over a five year period, you would easily lose, you know, 75% of your users. Um, now, you know, testament to a good application is how sticky are they? Right. Are you, are you. The ones who buy the next version.

Chris Gammell: Right.

Matt Berggren: Exactly. Decided. Now, what, what changes with subscription a little bit, and I know that, you know, this is a controversial topic, but I, I'm, I'm, I'm not unwilling to, to, to chat about what changes with subscription is that the relationship is now one in which I have to constantly, uh, work to convince somebody, not every year, but every month that it's worth spending more money with me.

Chris Gammell: Yeah. Yeah.

Matt Berggren: And every month, some percentage of those people make a decision that they're going to renew. And I need to be there with, uh, answers to the things that are frustrating them and the things that, that make them unhappy, make them, you know, uh, uncomfortable. You know, it's, it's a bit like people who have their, you know, $6 a month GitHub account so they can have more private repos, or I don't know if it's still $6, but it was $6.

Chris Gammell: It's actually free now.

Matt Berggren: Um, but yeah. Is it? Yeah. Thanks. Thanks. Thanks. Even Microsoft. Yeah. Right. Thanks. Uh, so, so, um, you know, it's, it's those accounts that, you know, you're paying something on a, on a regular basis. Uh, it, it creates a predictable stream of revenue, which I think is really important. Um, you know, there, there were too many years where it was feast or famine, um, because the amount of software engineering that you're putting into these things to solve some of these big problems, these are long lead time problems. Some of them are really, really hard to solve. Yeah. Structural stuff that will propagate.

Matt Berggren: Foundational stuff. Yeah.

Chris Gammell: Right. Yeah. I mean, I remember. You decide to rewire something and it's just like, oh my God, everything, everything touches that one piece of the graphic, you know, graphic thing. Yeah. Exactly.

Matt Berggren: Exactly. So, you know, probably the best case in point was, uh, the move to direct acts, uh, at, at Altium, which paved the way for 3d visualization, um, and the move away from, uh, OpenGL GDI type, type rendering to, uh, to direct acts. And that was because direct acts nine was the first to introduce instancing. Um, and instancing, if, if you don't know instance variables are a way to represent a single variable in memory and then instance it all over the place, uh, inside of the graphics pipeline. And, and, um, and that reduces the memory footprint and a lot of the computational overhead. So, um, anyway, the, the, um, the point is, is that you make a foundational change like that. It two years later, you deliver something like, you know, 3d visualization and users say, oh, well now we understand why you did it before we hate you. And those are long lead time problems. In some cases you're going to break things. You're going to go back in terms of capability, but you've got to have a sense of humor about that. And, uh, or if nothing else, you got to have a thick skin. Right. And, and I think the more that you're transparent with your users and you say, hey, you know, we're, we're doing this, but this is what we're aspiring to do the better. Right. Yeah.

Chris Gammell: That's, that's a tough balance to strike too. Cause you know, like I think a lot of businesses aren't prepared to say we're going here cause either don't want their competitors to see it or, you know, they don't know, they don't know if it's going to work. And so that's tough when you aren't able to say, hey, we're going towards this thing. Like I don't think you were allowed to say we're moving towards fusion electronics, but. Right. Right.

Matt Berggren: And we, we'd very deliberately held off on that. Um, and, and in part because there's, there's also competing financial pressures, right? Sure. Of course. Like, uh. You know, when you.

Chris Gammell: It's the, it's the cannibalization problem, right?

Matt Berggren: Yeah. Well, and it's, it's also a, uh, just a revenue problem, right? Like if, if people are, uh, buying because you've given them a fixed timeline for when something is going to happen and then you don't deliver. Oh, sure. Um, then, you know, can you really recognize that revenue as, uh, as tied to the product that you had at that point in time? So we've been very, very hesitant to say anything related to, you know, when this is coming out. And I, and I won't say that, uh, because there's. No, we're not getting a date today. I didn't, I didn't realize that. No, no, no, no. No. Sorry. Sorry. Yeah. But it's, you know, the, the fact that I'm showing it to you on screen gives you kind of an indication of where we're at. Yeah. Yeah. There's still, you know, quite a bit of work to, to be done between now and the time in which we release things. But there's some really interesting things that I think over the lifespan of a career were things that I kind of always hoped we would achieve, but, um, but were just impossible. And, you know, if you, if you think about mechanical design and manufacturing as being the core competency, certainly of my division at Autodesk and Autodesk owns Maya and Max media and entertainment division, totally different. Uh, and then, you know, AutoCAD and Revit and BIM 360, you know, these big, uh, uh, architecture, engineering, construction, uh, division. I live in the manufacturing division and in the manufacturing division, we focus on product design and manufacturing, right? So, uh, inventor is one example of that fusion 360 mechanical design and manufacturing, but it's, you know, cutting metal, 3d printing injection, molding with mold flow. It's, it's nesting, it, you know, sheet metal nesting, all of these sorts of things. So when you look at, at, uh, an industry like mechanical design and manufacturing, it's really, really only a couple of adjacencies, uh, you know, other industries that make sense, um, adjacent to that. One of which would be, you know, PDM PLM. And that's why I think a lot of the mechanical design companies, uh, have their own respective solutions. Fusion has fusion life cycle, which is, you know, life cycle management, um, and super powerful life cycle management that, you know, can, can store the, the state of a design as it moves through the pipeline. Right. And then on the other side of that is really electronics because there isn't a circuit board that doesn't get put into an enclosure. And if every enclosure was a din rail case, um, you know, all the electrical engineers in the world would be happy. Yes, I would be. I would be ecstatic. And nobody else would pull the blister pack off of the shelf. That's right. God, that's so pragmatic. Finally a place for me to put all of my electronics.

Chris Gammell: If we could just get like a din rail that attaches to your wrist, then we could finally make a good watch, you know?

Matt Berggren: Exactly. Exactly. Hey, you know, what is a Pip-Boy except for a din rail on your, on your wrist? That's right. So, so, so, so the, you know, the, the key here was we didn't want to just stick electronics design into fusion. So a few things that were kind of table stakes for what we wanted to do. Um, the mechanical world has things like thermal simulation, things like, uh, stress analysis, um, really good, uh, tools for grabbing a face and saying, generate a geometry from that or project a sketch from that. Um, the mechanical world has things which are totally missing in the PCB world, like spline curves. Oh, yeah. And, uh, and so I'm proud to say, uh, at least as far as I know, um, and so that's a pretty horrible qualifier, but, uh, that we've added spline curves for board shape definitions. Uh, and so that way we can now align to what the mechanical engineer is doing. So it isn't this piecewise approximation of a curve that's, you know, within marginal tolerances of what was going to show up in the mechanical side, but that those two things represent the same piece of data.

Chris Gammell: Yeah.

Matt Berggren: Moving back and forth.

Chris Gammell: So let's, let's do a little bit of structural stuff here though, too. So, so fusion electronics, is that what it's called? Is that the title? Or does it have a fusion 360? Fusion 360, but the electronics portion of fusion 360. Yep. This is now, so diverging from Eagle, is that right? Uh, yeah. So, so maybe, maybe dovetailing again later, but, but as of right now.

Matt Berggren: You know, I don't think we have to, so, so we don't have to kill desktop Eagle. Sure. And we have no intent to do that. Uh, you know, what, what people know and like about desktop Eagle as something that's just simple, easy to use, kind of hanging out on the desktop, you know, like that doesn't go away. But on the other side of that is the, uh, the, the sort of holistic electromechanical design and manufacturing. So, so, you know, bolting on to mechanical design and manufacturing, the electromechanical side of that. Yeah. That's really the, you know, that's the evolution of it.

Chris Gammell: Yeah. Well, I think about the, uh, so there, there was a converter at one point that, or that exists right now, rather, that allows you to go between Eagle and fusion 360. And that was a cool development, but that is basically a database transformation. It was, I saw it, it was like a click a button, things update, things calculate in the back end, and then you have the update over on one. That is no longer the case. Now it's basically you're in mechanical, you're working on electronics. Everything is just the videos I've seen from the demo was seamless.

Matt Berggren: Yeah. Yeah. And so, so there's, there are a few fundamental things which are important to understand. And I'll, I'll, I'll share with you, uh, because I'm sharing my screen with Chris, but I'll, I'll share with you by my screen, but then I'm also going to explain them as I do them. You know, fundamentally the, the major shift in mechanical design in the last 20 years has been parametric editing, parametric modeling.

Chris Gammell: Yeah.

Matt Berggren: So everything exists on a timeline. There's a continuum. Everything downstream is calculated by events that occur upstream. And so, you know, if I create a sketch and I modify something inside of a sketch and I draw, you know, some body, you know, I draw a circle or something like that. I grabbed the sketch. I say, okay, my sketch is done. I choose to extrude something. I extrude something by some height. I then shell something. Right.

Chris Gammell: And actually to take it back to what we were talking about earlier, this is like the Tinkercad thing. This is the recipe you had mentioned with Tinkercad is draw a thing, pull a thing, but these actions can be replicated arbitrarily in any software effectively. Right. Exactly.

Matt Berggren: Exactly. And now, you know, if I take something like, like what I've got on screen here and I've got, you know, a timeline showing up the origin sketch for something that I extruded and shelled can be edited. Mm-hmm. And in just editing that sketch and changing whatever that sketch is, I can recalculate certain things downstream. So, I can take something else and merge two bodies together with, or two sketch bodies together with one another and merge that thing into one thing and extrude it and create something. Right. The challenge historically with ECAD MCAD synchronization was that you would pass a static model, what we were sort of in the mechanical world referred to as a direct model, right? Something that doesn't have history, something that's non-parametric.

Chris Gammell: Right. It's almost a match. I mean, effectively, right?

Matt Berggren: Yeah, exactly. I mean, it's a relatively unintelligent sort of thing. So, you would grab the, you know, static model from somebody and you would say, okay, I'm going to bring this into the system and then I'm going to build an enclosure around it. Or somebody would build an enclosure and send you a DXF, right? Mm-hmm. Yep. So, STEP, you know, IDF, any of those inter-exchange formats, they are sort of predicated on the notion that I send you a geometry and you use a geometry. But the mechanical world, there is this real implication, which is that I have a timeline. And that timeline is fed by that geometric information such that the downstream calculation of whatever that thing is, it needs to be considered in the overall analysis. So, what we did with the Eagle to Fusion was we insert the PCB into the timeline. Okay. So, we create a sketch and then we insert that sketch into the timeline and then we extreme that sketch to create the board shape. And then from there, that board shape could be projected.

Chris Gammell: This is the current conversion or this is Fusion? Yeah, that's a current conversion.

Matt Berggren: So, that's a current sort of Eagle to Fusion. Got it. Got it. So, we insert that sketch into a timeline, we create a PCB, we place a component. So, fundamentally, ECAD, MCAD synchronization, the two things historically that we cared about were board shape and component positions. That's right. Yeah, yeah.

Chris Gammell: Okay.

Matt Berggren: Now, I would argue that that's a way too limited view of the problem, but, you know, I can demonstrate why that makes sense, why what I'm suggesting actually makes sense. So, you know, I can take now in Fusion, in the current implementation, and go the other direction, say I have a geometry. And from the geometry, like, you know, an extruded enclosure or something like that, I can grab a face, project something, and produce a sketch from some geometry somewhere. And on the basis of that, I can create a PCB. And now the geometry and the PCB are linked to one another in a way in which I have a PCB board shape that now matches the geometry that was used to create the enclosure. So, all of a sudden, then I have a PCB, which now is linked to some underlying geometry through associativity. So, if the PCB board shape changes, so I go from a 3D board to a 2D board, I generate a 2D PCB from the 3D PCB, and I start to manipulate the 2D board shape. I can cascade that change then back through to the 3D PCB, and those things just work forward and backward however I want those things to work. And that level of associativity and timeline sensitivity means that the joints that people create, the projections that they create, aren't broken as a part of that process. That's hugely expensive for the mechanical engineer. Sending them a new step model means all the downstream joints need to be reassociated. You've got all of this work that you've got to do. It could be hours of work, depending on the nature of the design. So, what we've done within the fusion environment is kind of take it a step further and say, can I take any geometry anywhere, it doesn't matter what it is, and project a sketch from it? And then from that, can I associate that sketch with a circuit board? And then from that, can I generate that enclosure from that circuit board geometry? And the answer is yes. Like, we can manage both of those things in a pretty clean, pretty tidy way.

Chris Gammell: So, is the assumption that the changes will be often enough that that will impact this kind of back and forth? Because I definitely agree that there's stress at the point of transition, right? So, I've handed off designs to mechanical as a step file. They've handed me back a DXF and say, no, no, no, the hole should be here. You'd redo this. Usually, it's at the beginning. But, you know, any time there's a transition of knowledge or data like this, it is a stressful thing, right? So, is the assumption that this is going to happen more often in this scenario?

Matt Berggren: So, I think the assumption is that it can happen more often. I would hesitate to say that it will. I think we are more deliberate today because of the lack of frequency in terms of, or the price of frequency, rather. So, you know, I make lots and lots of commits to a Git repo before I submit a pull request.

Chris Gammell: Right. Yep.

Matt Berggren: And I do that because I have the flexibility of committing something to somewhere other than my machine.

Chris Gammell: Right. Well, and also because the cost of the integration is high, right? Like whoever's doing the request merging, that's their time and you want to respect that.

Matt Berggren: Correct. Correct. And incrementally, people can grab my branch and they can test individual things. And, you know, I can give something to QA long before it's done, you know, air quotes around done. Right. But long before it's like truly complete and merged into the software and say, hey, just follow this branch over here. Oh, you didn't find, oh, I left the via button out of the toolbar or something like that. Okay, cool. I added it back here. Just pull this. And it gives us that sort of frequency in which we can interact as an organization. It's not exclusively the mechanical engineer, electrical engineer. It's kind of every other stakeholder in the process. So, you know, some of the capabilities within Fusion, like the T-Splines workspace, that's a mesh editor. And T-Splines was integrated into Fusion years back. That's a mesh editor for industrial design largely. So it's all of the complex contours and forms and sweeps and, you know, all the sorts of things that make a product look amazing. Get somebody to say, mommy, mommy, give me one of those and grab a blister pack. Right? Yeah.

Chris Gammell: So this is the one where you can actually like grab, it's like you're basically like molding in Fusion, you're saying? Yeah. Exactly. Exactly.

Matt Berggren: It's like kind of that sculpting style approach.

Speaker ?: Yeah.

Matt Berggren: Sculpting. That's it. Yeah.

Chris Gammell: Okay.

Matt Berggren: And that's necessary for really good consumer products, really good industrial design. Now, the mesh to B-Rep conversion is, okay, how do I make that then useful in the mechanical world too? Right. So, you know, in the mechanical world, I want solid bodies. I want to know that I can grab that edge and chamfer it. If it's a bunch of triangles, is that an edge? Right. I want to build it out with something. It's a lot of computation. Yeah, exactly. Yeah. So, you know, some of the things which have been really essential, at least from my vantage point, some of the things that I felt were really essential were that if we were going to do the 3D PCB capabilities in Fusion, that we didn't shirk on our responsibility to do things like question whether or not a board could be fully extruded, meaning that the copper could be extruded, the solder mass could be extruded, that I could have a positive extrusion for the copper along the Z-axis and then a negative for the solder mass, that I could have real solid body geometry. And there's a meaningful reason for that. When you go into something like thermal analysis, the thermal analysis tools, whatever it is, are going to mesh it for thermal analysis, right? It's FEA, right? So, they've got to triangulate that in a way that computationally works for thermal analysis. The CAM tools, the milling tools that would do something like allow you to cut a prototype on an LPKF or a Roland or another mill or any of those things, those tools are going to be a also will generally mesh the model when you go through the process. They mesh things slightly differently and it's opportunistic. If I want to 3D print something, it's much easier to start with a solid body and then bring it into the 3D print environment and print it so that I can do things like a test fit for connectors or make sure that the holes are in the right location and that my M3 screws are going to fit through that. I had the tolerance a little tight. I don't know that that's really going to work. So, what we did, which I think is really the more kind of revolutionary side of things, is we figured out how to do it all as solid body.

Chris Gammell: And so, can you contrast that to how that's done now though? So, solid body versus blank.

Matt Berggren: Versus mesh. So, probably an example, one example would be that, and this is kind of a more philosophical analogy, I guess. One example would be sort of the difference between looking at a picture of the Grand Canyon and standing at the Grand Canyon, right? Looking at it and actually seeing something. You've got this level of fidelity in the geometry. That means that when I highlight something, it doesn't just exist on the graphics card. It doesn't just exist in the render pipeline. It's a real geometry that can take a material and have the material applied to it. And it has this cascade of implications. Like, I can mesh it for electronics cooling or thermal simulation. I can mesh it for...

Chris Gammell: So, real quick on that then. Yeah. So, the idea would be if I'm drawing a trace now, it's effectively like I'm drawing a tiny square or rectangle. And then I'm extruding that along the path that I'm dragging with my cursor as a trace.

Matt Berggren: Yeah. So, that's pretty much it. We still want to keep PCB routing in the 2D domain, at least for now. I think, you know, there's all sorts of questions about whether or not we could go another direction with that. So, and this one didn't have the bodies loaded. But what you see here are kind of the breakdown of things. So, if I sort of remove the solder mask on a board, and I'm showing Chris the beagle bone. If I remove the solder mask on this, you can see like all of the copper geometry now can be highlighted. Yeah, prototype. It's not just sitting on the renderer. It's not just sitting on the renderer. There's real hit tests. I can really follow tracks in three-dimensional space. I can highlight individual signals and kind of follow them through the pipeline. I can see the individual signals that connect to individual component pads. And I can follow from component pads the entire net through a board. Now, just doing that was historically very, very difficult to do. So, computationally way too intensive. And so, in large part, people punt on this. But we figured out how to do it. And to do that in a way that's performant, where the copper is actually extruded, where it has real geometry to it. And then can immediately come into the thermal simulation environment with the properties of metal attached to it.

Chris Gammell: Yeah. So, to give a word picture for people here. Matt's showing me copper with a bunch of vias in it. And the vias are connected thermally to a ground pour there. But then I can see the segments that are drawing. You know, like when you see that kind of bullseye looking thing around a via when it's connected thermally, I see the actual circular segments around the edge, which it looks like is crafted versus, you know, however that would be done otherwise.

Matt Berggren: Yeah. And so, you know, in a mesh modeler, you get some high performance out of it because it's really existing in the rendering pipeline. In this case, we're treating this as real copper geometry. And as a result, we can apply materials to it. And if I can apply a real materials definition, and I don't mean an appearance, there's a distinction there, which is important because materials have reflective properties and all sorts of thermal characteristics. Yeah.

Chris Gammell: Not just graphical stuff either, right? I mean, you have-

Matt Berggren: Exactly. And then to hit that with, you know, 30 frames a second tumbling, you know, spin in a view cube somewhere, and you lose a lot of the fidelity when you look at something over zoom or something like that. Right. But to hit, you know, 30 frames a second or 40 frames a second or 60 frames a second or whatever, it's non-trivial. But that solid body copper means that we're introducing potentially tens of thousands of edges and we still have the level of performance. That has never been done before.

Chris Gammell: The edges, can you explain the edges piece? So the edges-

Matt Berggren: Yeah. So the edges are the actual edges of the solid body. So, you know, if you think about each thing having an edge, in a triangulated mesh, everything is triangulated. There is no edge. It's just triangles. The triangles have their own respective edges, but there is no shared edge across those things because it's been meshed.

Chris Gammell: So this is just math in a processor, math in a GPU kind of problem?

Matt Berggren: Correct. Correct. But a really sophisticated kind of problem about where do you put it in the nodes in the tree and so on and so forth. But then, you know, doing the equivalent with the solder mask to say, okay, now I'm going to do a negative extrude of the solder mask on top of that doubles, compounds the problem, right? Now you've got twice the geometry. Right. And then to layer components into that and still have it be performant. And then doing that with a pipeline that is still parametric is a hugely complex computational problem. But we nailed it. We achieved it.

Chris Gammell: That's great. Can you speak a little bit about the benefits of what will come after this? So like thermal simulations and high-speed simulations?

Matt Berggren: Yeah. So thermal simulation is one that jumps out as the most obvious in the short term because fusion already has thermal simulation. So my ideal case for thermal simulation, we're still a little ways from this, but my ideal case for thermal simulation would be you drop a QFN on top of the board and you put your vias underneath that QFN faithfully hoping that you're going to wick away the heat from that device and transfer it to the opposite side of the board. In that case, how do you know whether you have the right number of vias, right? Oh, sure, sure. You know, is that the right combination, the right position? If you follow the data sheet, you hope that that's accurate. But did the data sheet tell you that they used one ounce copper, half ounce copper, two ounce copper, maybe they did, maybe they didn't.

Chris Gammell: Did they simulate it? Did they actually just do it empirically?

Matt Berggren: Did they just, yeah, exactly. You know, hey, you know, we did it and it's not catching fire. Right, right. You know, so every customer who wants to build a consumer product, for example, is going to go through underwriters. Part of that effort going through underwriters is going to be that you need to account for those scenarios. Let's say I'm building a baby monitor, right? Wireless baby monitor is probably one of the more disastrous things for anybody to endeavor to build because you've got FCC certification, CE certification, underwriters, you know, you've got all sorts of stuff there. Then you've got the security problem of you're transferring video over of your baby over a wireless connection. So you've got the security issue. You know, there's just a whole bunch of problems that. So, you know, do you know that it's going to be 50 degrees C on the surface? So what this paves the way for, and this is, you know, I don't want to sell snake oil futures, but what it paves the way for is being able to look at some of those problems and then evaluate them in the totality of the enclosure to say the temperature rise of this part will drive this property above this threshold such that the material itself that represents the enclosure, whether it's, you know, ABS plastic being injection molded or whatnot, will transfer this much heat to the other side of that. And that means that the surface temperature of this at this point is X. That's the, that's, that's almost the Holy grail. The real Holy grail is that at some point you feed that into a pipeline that can then say, okay, well, what happens if I move it over here? And what happens if I move it over here and what happens if I move it over here?

Chris Gammell: And so, you know, some of the, like a generative kind of stuff too, then.

Matt Berggren: Yeah. So some of the, some of the stuff that we've been playing around with and there's quite a bit kind of in flight right now, but some of the stuff that we've been playing around with are things like, you know, we added the quick route capability to Eagle a couple of years ago and that's sort of like one click, one kill route something. Um, you know, if I decided at some point that I wanted to, uh, you know, rip up and reroute apart, um, the, the, the one click, one kill of the quick route capabilities is, is not intended to be, you're all singing, all dancing auto router. Um, what instead it's intended to do is say, Hey, I've got some connection from here to here. Uh, just salt that for me and it can do different pairs and it can do length tuning and you can do a variety of different things. So, you know, but the, the, the, that was really paving the way for a reroute command to say, you know what, I, I want you to, and if I can remember how to spell it, so it's reroute device, uh, um, and it would go in and recalculate the connections on a given device on the board. And this is a fairly sophisticated board to try to solve that problem for, but it would go in and try to re rip up and reroute the connections that would connect through. So you can see it, you know, it made some changes to the actual connections of you to here. Uh, if I sort of undo those changes, you'll see that, you know, uh, it actually rerouted those. So the reason to do that was that now the move command in the mechanical world, which can be used to move connectors, which is largely what the mechanical engineer is going to move, um, move a connector around can have an option to automatically rip up and reroute the connection on the PCB side, on the 2d PCB side. And that paves the way for us to have some really, really interesting workflows that then eventually tie into, uh, the generative stuff. Now move this part. Okay. And based on that, I want you to, uh, rip up and reroute this one or rip up and reroute that one, uh, based on the changes that I'm making to these individual things. And it, you know, that's when it starts to get really, really exciting. So what we've done is managed to kind of pave the way for all.

Chris Gammell: Yeah. That was the demo video that I saw for, um, uh, that you, that you just shown off at the Australia, uh, conference. Yeah.

Matt Berggren: Yeah. Yeah. Yeah. So, you know, there's, there's a whole variety of those sorts of things that, um, are in the works, uh, and, and being kind of, you know, thought through in a very, uh, I opened up a blank PCB being thought through in a very kind of intelligent and, uh, well, hopefully intelligent. I, you know, I'll, I'll let, uh, history be, be the judge, but, um, but being thought through in a way that has a real awareness of the overall pipeline.

Chris Gammell: So this is, this is actually really, this is great. Uh, this is great stuff. And it seems like you're, you're looking for feedback on all this, uh, where, where can people find out more, maybe watch that video, the demo video, and then see more about this and get in contact.

Matt Berggren: Yeah. So, so you can follow me on Twitter. It's Technolomaniac. Um, you know, we can throw it in show notes or whatnot, but, um, I've been, you know, they say the ship leaks from the top, right? So, you know, I've been leaking out information on a pretty regular basis. Um, but, uh, you can follow me there. There'll be a bunch of stuff. You can also see the Autodesk Eagle and the Autodesk Fusion 360, uh, Twitter feeds, uh, composing a bunch of videos now, and I'll share, you know, a bunch of that content and get a lot of that stuff out there. Um, and then, you know, as, as we get closer to the release, um, you know, we'll, we'll definitely let people know and try to bring, bring in a community of people to really test things out and help us to, to make the right, uh, the right decisions prior to, to putting this all, uh, out there. Awesome.

Chris Gammell: Well, this has been a whirlwind through, you know, EDA, you know, I really do. I mean, like, you know, I, I'm, I'm sorry. I'm not switching over from Kike at any time soon, but I am a Fusion 360 user. I, I love the software. I think it's, you know, I think it's been making lots of great strides. And I think all of this stuff is a really, um, you know, interesting seeing where things are going. Um, it's, it's very interesting, man.

Matt Berggren: It's, you know, it's a bit of a magnum opus, so we'll see how things go. You know, I, I, I don't want to mic drop the mic and walk off stage. I think what we've done right now is really race, you know, to, to sort of mix metaphors, you know, is we've raced to the start line. So we're, we're at the starting line now.

Chris Gammell: Yeah. It's a new paradigm.

Matt Berggren: Our battle to win, you know?

Chris Gammell: Yeah, exactly. And I think that you're going to, you know, it's going to be a little bit of a struggle to get people to try new things, but even Fusion was a, you know, a new way of getting people to do mechanical CAD. I think that, you know, there's a, there's a lot to be gained from, you know, kind of meshing of, of the software and mechanical and electrical worlds. And I think there's going to be a lot of interesting things there.

Matt Berggren: Yeah.

Chris Gammell: Yeah.

Matt Berggren: I agree. I agree.

Chris Gammell: Well, I'm glad, glad you have been doing a lot of cool stuff, Matt and keep up the good work.

Matt Berggren: Thanks man. Thanks. I really appreciate it. And thanks for having me on. It's great chatting with you. Yeah. Thank you. All right. Take care.

Speaker ?: Thank you.