Episode #103 | Design for Medical 3D Technology (Virtual Event)

Healthcare 3D printing is moving fast, and design is leading the way. In this episode, we explore how advanced CAD, simulation, and automation are enabling patient-specific implants, multi-material tissue-like structures, AI-powered prosthetics, and fully custom pediatric seating. Beyond the printer, human-centered design and smart workflows are turning ideas into devices that improve patient care. 

We start with the biology. Orthopedic engineer Matthew Shomper of Not a Robot Engineering, LatticeRobot, and Allumin8 explains why stress shielding sets up decades of problems and shows how patient-specific scaffolds can be generated in minutes. Analyze intact versus defect states, compare strain fields, and synthesize a topology- and strain-matched lattice tuned to a person’s real loading. Swap patterns, change valency, target grafting, and even plan for resorbable polymers as bone fills in. It is a shift from “stronger” to “more biologically honest.”

Then we open the toolbox. With volumetric and implicit design approaches explored by Rob MacCurdy at the University of Colorado Boulder’s Matter Assembly Computation Lab, design moves from surfaces to functions that define geometry, material, and behavior together. Think functional grading across a dogbone, gyroids blended between materials, or lattice struts whose composition varies along their length to steer buckling. The same logic can drive multiple printers and processes, enabling surgical models and tissue-like parts that span from soft to structural in a single build.

The payoff comes at the point of care. In prosthetics, comfort is the foundation. Joshua Steer, Founder and CEO of Radii Devices, shows how data-driven rectification gives clinicians an informed starting point they can refine. Nathan Shirley of HP explains how automation turns that interface into a robust, production-ready socket with a single request. No brittle CAD models. No days in design. And in pediatric seating, Alexander Geht of Testa-Seat shows how lightweight, water-cleanable, fully custom supports help children eat with family, attend school, and travel without a van full of gear.

Validation, reimbursement, and regulation still lag behind what is technically possible. But with open toolchains, integrated simulation, and outcomes data, patient-specific devices are moving from heroic one-offs to dependable care. Subscribe, share this with a clinician or engineer who should hear it, and tell us the one custom device you wish existed. What would you build next?

Video On Demand 

Chapters

• 1:07 Welcome, Mission, And Community
• 1:28 Why Design For 3D Printing In Healthcare
• 2:30 Orthopedic Implants And Bone Mechanics
• 6:18 Stress Shielding And Aseptic Loosening
• 10:40 Patient-Specific Scaffolds And Strain Matching
• 14:35 Live Workflow: Fast Personalized Lattices
• 17:13 Tooling, Validation, And Regulatory Realities
• 19:45 OpenVCAD: Volumetric, Multimaterial Design
• 26:05 Functionally Graded Materials And Lattices
• 31:20 From Code To Printers: Outputs And Slicers
• 36:04 Medical Models And Tissue-Like Mechanics
• 40:18 Access, Licensing, And Standardization
• 45:28 HP x Radii: Scaling Better Prosthetic Fit
• 51:08 Clinical Loading Zones And Socket Design
• 55:20 Data-Driven Fitting And Hybrid Strategies
• 1:00:12 One-Click Socket: Design Automation
• 1:06:08 Computational Workflow Deep Dive
• 1:12:08 Custom Pediatric Seating With 3D Printing
• 1:18:40 Configurators, Portability, And Daily Life
• 1:24:10 Reimbursement, Regulation, And Scale
• 1:29:05 Audience Q&A: Tools, Training, And AI
• 1:38:00 Closing And Upcoming Events

Transcript

SPEAKER_01: 0:00

Okay. Good morning everybody. Uh I'm hoping this they uh Zoom changed danger phase again since last time. Thank you very much, Zoom. Um just want to have a quick intro about 3D Hills. We have started this company about 10 years ago with three missions. One is to educate the public about what 3D printing and 3D technologies can do for healthcare in the life science space. And two is networking. Um, we have now a combination of in-person and hype uh and uh virtual events to connect people. So for virtual events, we suggest that you share in the chat box who you are, your social media link in case you know you want to meet your kindred spirit. And we have an in-person event coming up specifically for startup founders and investors, accredited investors in San Francisco on January 11th, right before JP Morgan. So I will share a link of the event. And if you qualify for either, please join. It will be free for either startup founders and investors. And the other people, they they may be able to register, but it will go through approval process. And number three is related to events is to help early stage startups with fundraising. So early stage state is defined as pre seed, which means almost like you have a prototype, but really kind of looking like a rough, rough sketch in reality to Series A. Um, so if you qualify, let me know, email me. Um okay, well without further ado, I'd like to uh introduce the topic today, which is designed for 3D printing and 3D technology. It's a broad topic because uh I want to keep it broad so that people can have freedom to operate. And the first speaker is Matthew Schamper, which a lot of you guys know because he shares amazing posts on social media. And today we finally got the opportunity to see him in in real life, although virtually. Um so Matthew, I will let you take it away.

SPEAKER_05: 2:08

All right, thank you very much. Uh I apologize, I've got a little bit of hoarseness going on. So um if you can't hear me, then uh I'll try to speak up my wife's head. So yeah, I'm not gonna go in through through an intro for me. Uh I'm assuming everybody can see my screen. Um so I'm gonna go through a little bit of a this is kind of kind of a fun one for me. I've got 15 minutes, so I'm gonna go through a little bit of slides, not a ton. We'll probably bruise through them because what I want to show you is the design side, you know, focus more on the design side of additive uh 3D printing designed for um healthcare in particular, uh, with my background has been virtually all in orthopedic implants. Um and so when I when I think about the evolution of additive manufacturing and the software in particular, um I, you know, my mind goes towards how do we how do we leverage its use for healthcare, um especially now that things like 3D scanning and imaging techniques, um, additive manufacturing techniques, processes, things like that, biocompatible materials, bioprinting, as a lot of these things have come into the forefront of development and things have gotten uh approved by regulatory bodies, it's really exciting to be at the intersection of uh design, manufacturing, and then application. So those of you guys have heard me speak before may have seen similar-ish slides before. One thing I really want to target, though, is how this these new technologies are enabling us to create targeted therapeutic uh um solutions for individuals. So less so a broad brushstroke, um, you know, not like a one size fits all, but a specific size fitting an individual person, which I believe is the next generation of healthcare products uh that are targeted for an individual in their particular condition. So uh orthopedic products are essentially products that interface with the musculoskeletal system, um, bone in particular. And bone is a living vascularized uh organ, right? The they are um it's able to heal itself, it responds to stimuli and stresses. In particular, this is one of my favorite graphs. There are there are zones in which bone is loaded that have different responses. And of course, there's a really complicated biocascade that happens, but in general, uh there's a range of microstrains that cause bone to grow faster. When you have an issue, fracture, um, fusion because of some uh nerve condition or something more complicated. Generally you would like to target a particular strain range, which helps bone grow faster. Um there are various principles to how you fix fuse bone. Um I would encourage you guys to look those up. If you're not familiar, the AO principles of fusion are what I like to go to. They're four or five, depending on which website you go to, of things that we try to do. Um when we load bone, uh, or when we're trying to fix bone, right now we do it poorly. And by what I mean is we're we're, you know, our best options are chunks of metal that we put inside bone. Um so here's showing a hip stem being placed inside the bone. What happens is you see in the intact, we've got uh stresses and therefore strains happening on the exterior of the bone. Um when we put the metal in, the metal is now taking up a lot of the loading. Uh, and we're removing that loading from the outside of the bone. What we've seen in studies is that cyclic bone stress changes as little as 1% could have a profound difference on bone remodeling. What happens is that stress becomes shielded by the metal that's in there, and then we start to see bone resorption. Going back to this graph, when we see strain below a certain rate, the the bone says, I don't need you anymore. I'm gonna harvest you for materials because you know it's looking for homeostasis, a balance. So the problem is uh, you know, we start to see stress being distributed away from the cortical bone on the outside, which leads towards massive and profound bone loss. Um the resorption areas is again fairly profound. You can look at several studies which show virtually non-existent bone towards the top half here, which causes a lot of issues. So, you know, what do we when we try to place implants, what are we trying to prevent? Well, one is infection, but infection is actually pretty low incidence, um, you know, typically between 1 and 3%. But what is a much higher incidence rate is what's called aseptic loosening. Uh it's basically where um through multiple different factors, the implant doesn't sit like it's supposed to, or over time it starts to loosen, causing additional degeneration, uh movement of the implant, um, subsidence, which is the implant subsiding into the weaker bone, uh pushing into it. And this actually occurs very commonly. Um aseptic loosening happens in over 50% of hip uh revisions, which is significant. I mean, it's it's kind of the the dirty secret of orthopedics is that you know we put things inside the body and attempt to fix it, but the stiffness uh so drastically doesn't match the bone that we we end up with further problems. One of the other key words that a lot of people like to talk about is adjacent level uh issues. So basically we fuse a segment. It's all nice and stiff now with a piece of metal in it. Well, everything that happens after, you know, on the outside of the bone now sees different different loading patterns. So now we see a cascade to adjacent levels. Um 33% of people who have a low back fusion will end up with SI joint pain. Your sacroiliac joint is essentially your hip joints, um, especially in women. So we're essentially just cascading the problem outside. I think, you know, again, of the aseptic loosening factors, micromotion, patient factors, surgical technique, some of these things um can be mitigated through better techniques. Uh patient factors are tough. If a patient is obese or a smoker, they're we're gonna have challenges in treating them. Uh micromotion is uh is something we can solve with 3D printing because we can add texture to it, let's say. Or implant design, the shape and the materials can affect its stability and its stiffness. So one thing I like to focus on is this implant design, because it is a direct measurable thing that we can affect, especially with the advent of 3D printing. So, you know, this is one of my favorite quotes. Um I I coined this, I don't know, probably close to a decade ago, started talking about this. And there's a lot of other people that talk about this. Zorqa Booser Bucer, who's at the hospital for special surgery, talks about this a lot as well. Um, but I I like to say the best implant is one that's forgotten by the body's biological systems. Meaning if it can if it can be a direct match for stiffness and therefore strain, uh, or if it could ideally resorb over time, which I believe is going to be a reality in the next decade or so, then we solve the problem of this stiff, mismatched, solid piece of metal living that person's body for the rest of their life and causing issues. So how do we do that? Can we maintain stiffness while strength increases? I think a big portion of it is to be able to start looking at implants as bespoke personalized products. Anymore now, and if you go and buy something, you want something personal to you, right? Um I want this with my initials on it. I want this color to match what I'm looking for. The problem is, you know, we're we're still in in implants, we're still doing um uh one a one size fits all, or a few hundred sizes that we happen to bring in a set has to fit this patient, the best fit. So the surgeons have to do a lot of sawing and trimming of the bone to get the implant to fit the way that they want it to. But I don't think that has to be the case. The other problem is that because of regulatory issues, we have to pass the FDA or other, you know, the EU's MDR requirements, we end up having to test to absurd limits compared to a clinically relevant biomechanical load. Here's one of them. Um 40 times difference. And this is again, you're pulling from direct or ortho load, you know, concatenated data from research papers uh versus uh a 50th percentile load required to pass the FDA's testing based on devices that have been cleared before. So we're talking about a 40x difference in strength and therefore stiffness, right? So how can we start to change the paradigm using 3D printing? So I I I'm gonna take a little bit of a risk. It looks like I've I've used about 10 minutes to uh uh to do to do some um some live modeling. I I I I love doing this. This is fun. So essentially what we have here is we've got uh a humerus. This was a bone in my repository that I had available. So even though we're talking about hip implants, we're using a humorous just for an example's sake. So what you'll see here is essentially this is, you know, we've got a we've got a bone and we've got a void that we need to fill. Now, this void is most likely because of some sort of uh oncology-related um thing, you know, some sort of tumor, some sort of resection that had to happen. Could be because of an accident, but we're just gonna presume there's a section of bone that needs to match topology. Um what we can do now is we can essentially set up analysis, right? So we're gonna set up essentially what this thing looks like, the strain distribution uh when this bone is loaded in a certain condition. Because we know if it's intact, we know how the bone should act, the strains that should be present in the bone. Then we can essentially perform a simulation on the voided bone, right, with the issue. We note its strain pattern. Um what we can do then, excuse me, is we can look at the various strains, and what we see here is that for intact bone, we can see when we load it in this manner, we're presuming this person's you know lifting some weights, right? A 50-pound curl essentially. They're right, they're in that deal range, right? 2,500 microstrain. Um however, when when we account for the void, we now see that various portions of this have pushed into that overuse where we're going to start seeing some issues. We also see that because of the distribution of stresses now, we're going to have some stresses which are falling below our ideal stress range. So, how can we sort of how can we sort of change this? Well, again, utilizing computational design, we can basically take this field and we can look at the difference between intact, what it should look like, and what it doesn't look like, that allows us to create a targeted scaffold that not only matches the topology of the bone, but also matches the strain range of the bone. And we've got lots of dials and things that we can pull, right? Levers and stuff that we can pull. But essentially what we're trying to do is we're trying to match the topology, and then we're also trying to match the strain, right? More important than topology is strain. And what we can do is, you know, again, relatively quickly, I think most people are surprised by how quickly something like this can happen. We can say, okay, well, new patient, new bone. Um, we're going to take the point uh that we're at, um, and we're gonna just pull it to a different spot, let's say. And now we've got a new bone with a different void. Again, the void's gonna look the same. I'm not changing anything. And then in uh, you know, with a full static simulation comparison, um, we can essentially get a targeted um solution in what was that, 30 seconds? Um, that's topology matched as well as strain matched. And, you know, even you know, even more similar, right now we're doing like a like a face-centered cubic design. But let's say we want to go with something that is um hexagonal. We can do that too. Uh, and again, in you know, very short period of time, we have a strain matched hexagonal pattern that's purposely built for that person's body. We can even change things like loading. Let's say it's somebody who is smaller or larger, they're gonna see different types of loads on that bone. We can target it, you know, with a long boat in the leg to a gate analysis. We know exactly what forces that person's going to see. Um and then, you know, even something like we can adjust structural offsets, we can increase connectivity. Uh, let's go ahead and change some valency points in order to increase or decrease the um the effect of the strain and how open or closed we want to make this for things like bone graft packing, or again, future considerations are something that's gonna resorb over time. We can make this match the stiffness for now of the bone. And then as bone grows in, we eventually resorb the scaffold over time, which is, you know, honestly, polymeric scaffolds um that are resorbable is a new area of science that is starting to see some traction in early stage research for patient stuff. So I think I'm at the end of my time limit. Uh, Jenny, I can give it back to you. But um, really, really neat stuff from a design standpoint that I think will um be sort of these upfront considerations as we start looking at these targeted patient-specific care and these sorts of tools and abilities would not have been even remotely possible a decade ago. And now that their software has has moved into the phase where we are able to sort of directly do this, it's it's really exciting. And I think it's a fun place to be in.

SPEAKER_01: 16:50

So yes, absolutely. Um, I just want to mention the software you use is Antopology. And we're not sponsored by Anthopology, but we welcome future sponsorship. However, um uh people want to use it. You know, I just want people to know that that is the software that you demonstrated. Now, my question is that desktop screenshot that you shared, was that your personalized setting for this particular patient, or is that like available for everyone already?

SPEAKER_05: 17:18

No, I mean the the short answer is a lot of these softwares, and you know, Nathan I think is really familiar. Like you basically are you build a workflow with IOs, ins and outs, and the workflow can be reusable, but it essentially has to host be hosted inside a software. So NTOP, what's really cool is you can essentially build almost like a Python-esque workflow through nested blocks, and then you can you can package it up, give it to a end stage user, and all they have to do is drag the bone in, set some of the conditions, and then they can essentially get the output that they want. So the the short answer is um, you know, what I'm currently working on isn't like an end stage commercial solution, but it's fairly easily programmable. And one thing I will mention, a little bit of a plug, is I'm actually going to be speaking at Rapid next year about about this topic expanded. We're actually going to be doing some additional research on if the design that we get actually matches the stiffness range of the bone. In other words, it's one thing to say computationally this makes sense. But we're going, you know, it's an essentially this effort, right? This it's like the you know, I think the title of the catchy title is in search of the perfect strain. And this has been something I've obsessed over the last decade of my career. Uh, is this idea of we know the strain, we can, we can model it, but how do we actually come up with a scaffold that purposely matches every bit of strain around every bit of the of the patient matched geometry? Is that even possible? I don't know. Right.

SPEAKER_01: 18:52

Um yeah, the other question is, you know, you've showing all the of the finite element analysis on it. Is that part of the package, or is that something added to it?

SPEAKER_05: 19:00

Aaron Powell It is. You know, I would say the caveat is we're looking at it's quasi-static. Um, I think that there would be some people, especially of like cyclic loading, would which would want to see uh some sort of uh nonlinear analysis. Um, however, I've done a ton of work in simplifying problems to quasi-static in order to make them a reusable workflow. So I think that you can get around it. But the short answer is yes, it's integrated into NTOP. They actually have meshless simulation um as well, which is incredibly fast. I mean, this took 30 seconds. Some of their meshless simulation tools that like this would have been solved in six or seven seconds. So you're talking about extraordinarily fast. You could go through scores of iterations and be done in a couple of minutes, let's say. So that you know, uh but you could also wire similar tools. So um Grasshopper, Houdini, these are other computational tools that can do. similar things to end top things for missioning yeah even like even like uh um oh shoot my my mind's blanking even like blender some people have started to use blender with various python scripting tools to do similar things and so d depending on where you're at with like your ability to to pay or not pay for a commercial piece of software there's lots of options I think to to get into the space and do something similar for sure. Yeah and also you know just because something visually or maybe even mathematically is valid like you mentioned you still need to validate it right in the in the especially for med especially for med tech yeah this is a this is a cool example to show uh you know what I'm assuming is a technical audience but to go from this to a validated solution I mean you're looking at you know animal studies patient trials potentially however like uh I can I can tell you that this type of work is already being done um I'll I'll give him a shout out I I I love him like a brother Dr. Oli Kiapoor he does a lot of work with a company called forweb um he does a lot of work with General Mass and Harvard and he's currently looking at similar basically a a a sort of a long bone implant that's that's patient matched and targeted.

SPEAKER_01: 21:15

We've done some work together this is he's working on this without me you know I'm only a little bit jealous but but there's active work right now that that's going on it was super helpful uh Matthew if you can put a link uh of him or whatever it's like a scholar or a Google scholar thing into the chat so that people can look him up. Okay. Sounds good. Um okay well thank you so much. Um everybody if you have questions for Matt or everybody uh who's speaking please put them in the QA box so I can keep track of it. Um I'm I'd like to introduce our next speaker who is Professor Robert McCurdy he is assistant professor in mechanical engineering and computer science at the University of Colorado Boulder and he has really some really cool projects and he leads the Matter Assembly computation lab.

SPEAKER_07: 22:05

Hi thanks for the introduction yeah I'm Rob McCurdy um and I wanted to tell you about one of the tools that we've made in my lab called Open VCAD we are a computational design and fabrication lab and so um this is a tool that we developed internally and now we're making it available for folks externally as well I'll tell you a little about that uh at the end. So OpenVCAD is what we call a multimaterial design compiler emphasis on multimaterial and I'll show you what I mean by compiler in a moment. I think the motivation here is that most design tools for CAD, you know, which also means 3D printing were predicated on the idea that you would model what was on the surface and then what was inside that surface would be the same thing whether it was one material or some composite you didn't you don't have the ability to control spatial variations. Yet there are increasing numbers of 3D printers today that can handle spatial variations internally. Those printers include multimaterial extrusion printers with material swapping and also dual extruder increasingly available VAT photopolmerization. This is still kind of in the realm of um I guess research curiosity but um there are increasing numbers of research results showing uh various flavors of multimaterial VAT photo inkjet for sure has existed for quite some time as a multimaterial option. Stratosysis printers have been doing this since 2025 sorry 2005 um there's also multimaterial powder bed in various flavors whether it's metal or polymer so you know I think that the the hardware has been there for quite some time it's actually the software tools that are lacking. And when you see folks out kind of at the research frontier who are doing multimaterial design and fab, they're usually writing their own codes to represent the design tools and sorry represent the designs and that's what we were doing in my lab and we kind of got tired of reinventing the wheel so we create this thing we're calling open VCAD. So in the interest of time I'll just skip past some of this I think this audience is probably pretty familiar with what I mean by bound representations and CAD open VCAD is a volumetric representation. So this is an implicit uh representation we use functions and operations with a directed acyclic graph this is informed by an old history of constructed solid geometry if you've heard about that. So you specify what your design will look like as a sequence of operations and functions and then you compile those operations and functions into an appropriate design file for whatever printer it is that you are targeting. We have recently standardized on Python so there's now a Python library in fact you can just get this from various package managers we've made the the bit available so people write their file in Python and then call additional tools in the VCAD library to do the compilation that gets compiled to whatever output format is necessary for that particular printer and away you go. So quick examples here if you want to create some object that looked kind of like this which is a two material object composed of both red and blue regions and it's also the difference between a cube and a sphere well you could you know call difference here on a cube and a sphere and then use something that we call functional grading or F grade. And this implementation and this usage it allows the user to say well I want 20% red and I want 80% blue to be randomly distributed throughout this object. There are a bunch of other ways that we can distribute material in an object and I'll show you a couple of those next not just random so go past this. Here's another example canonical dog bone if you want to have like a functionally graded dog bone with an interface between the regions. I'm showing by the way large blocks here to delineate you know one region from another but in reality these blocks would be very very small so you wouldn't see the the blending occurring so here's an example of using the functional grading block with an inverse normal distribution and a normal distribution. These are functions specified elsewhere in the description and then we're grading between material A and material B over in this case a part that was imported from another piece of the file this could be an STL this could be some implicitly defined object and then we just functionally grade over it to get to get this object so I'll go past this. Another example this is fabricated on a polyjet system here we're importing a screwdriver STL as a mesh and then saying that when X is less than 100 we're with some probability going to be clear versus pink versus yellow. So we've basically made the probability 100% that when it is less than 100 it's going to be all clear. And then when it's greater than 100 with inverse probability it transitions from yellow to pink in this case I should have reordered these but I think you get the idea. So we've got some simple functions here that can be compiled down during the um the compilation process to create spatial variations in the material. All right we um you know have uh worked in lattices everybody likes to work in lattices there's lots of cool applications for them um we can certainly work in lattices as well this is one way of working in lattices where you import a previously defined lattice and just functionally grade over it. If you can squeeze these you would feel them transitioning from incredibly stiff on this side to incredibly soft on this side we actually are uh I'll say this a little bit although it's not the focus of this talk um my lab has um been one of the early developers of the liquid printing for the polyjet systems and so it turns out that you can mix in a non-curing liquid which is the cleaning fluid in the polyjet system that changes the mechanical properties of the material pretty profoundly so we can get far softer material than would be possible with the softest uh base materials in the polyjet system and so this thing uh ranges from um tens of kilopascals all the way up to almost a gigapascal on the other side and you can really tangibly feel that when you manipulate it um you know there's lots of interest in in all kinds of meta materials we've been lately using OpenVCAD as our engine for representing metamaterial designs for other application spaces including impact mitigation and uh medical planning. One of the things that you might like to do when designing a metamaterial is to have lattices that incorporate additional complexity. So for example if you want to have spatial variation along the length of each individual strut in a lattice OpenVCAD is a straightforward way of doing that. In this case I just graded the uh density of red versus blue along the major axis of the lattice but you certainly could grade them from side to side as well. You could potentially do that and you know dictate the way that this thing would would bend or would buckle in response to an applied load. Here's one example of a simple body centered cubic uh that's being uh first functionally graded along the direction of the lattice so you can see that in the center it is uh one material and not toward the edges it's another material. We then uh have a you know like every CAD tool have a tile uh capability so we can then tile that design throughout um you know whatever space it is that we want um this is just kind of an you know example of a bunch of different things thrown together so this is a a gyroid that is bounded by a NACA airfoil. The gyroid itself is spatially uh graded in its extent so in other words the size of the gyroid that's straightforward to do with the gyroid functions what's not straightforward to do is to then blend from one material to another within the gyroid uh so VCAT is allowing us to do that and it's a really um quite uh simple design file even though it's a complicated structure I should have said that as well on that uh lattice that I was holding and squeezing um that's a lattice that's about 750 million voxels uh in a very sparse representation if you were storing that as a file it may maybe be like a billion and a half bytes and in open vCAT it's just a handful of lines three lines or so um I also think versus voxels implicits are a great way of passing information around um we can compile open vcat into a bunch of different output formats so for sure we can do visualization we can do uh voxel or bitmap printers like inkjet uh binderjet uh vat footle polymerization we can compile to extrusion based printers so direct inkwrite FFF uh etc we can also um export uh multimaterial meshes I'm not really talking about that much in this talk but we have a paper out this fall in added manufacturing uh which demonstrates exporting to um either level set like meshes or even um we have a slicer that we built as well so we can directly control the blending of multiple extrusion uh nozzle type uh printers if you want to have continuous gradients uh in a in a design that's based on extrusion. Um I'll go past some of this a little bit let's see yeah I'll go past some of this as well this is just kind of detailing how we go from a design that's uh truly continuous to a design that can be implemented on a variety of different printers. If you have like a a Pruscia XL, you've got five different materials and so here n would be five. If you have a a mixing material um n could be uh theoretically infinite our slicer can handle both of those um so we try to make it easier for people. Again it's possible for folks to do things like this by inserting their own G code snippets or modifying G code. Uh increasingly some of these slicers are becoming more more functional but still you have to have um a pretty significant understanding of the way the G code works if you want to implement designs like this on a mixing or a multimaterial extruder type printer and so uh VCAD is meant to abstract all of that complexity away from you. So I'll just move on. Yeah so this this tells a little bit about the ins and outs of VCAD so I'll dwell on this a bit. We certainly can import meshes so we can import uh 3MF and STL we can import CAD files uh you know standard step formats uh we can import FEA simulation results so this often takes the form of uh vector fields of deformations we can certainly import DICOM medical scan data we've got a couple of slides talking about some of our Fourier's into medical printing uh we can import implicit surfaces and we can import uh voxels if you have you know a voxel definition and then we can output output to uh bitmap stacks like uh the Polyjet uh voxel import utility would take or uh increasingly some uh VAT photopolymerization systems will take uh we can outport output to uh simulation files so um you can send a file that's already meshed and has the simulation parameters already defined as a dotimp for Abacus if you're a user of that um it's actually quite similar for some of the other popular FEA packages uh open and closed source um we can output meshes we can as I showed a little bit we can output G code we can directly output Vox gluids for later processing somewhere else and then of course we can render and view these uh results so um this is uh the initial paper for open vcad 2024 added manufacturing and uh there's uh another paper also in added manufacturing this fall which talks about the G code export capability um we have deprecated the VCAD studio in favor of Python implementation so um you can readily get this library using the Python package management tools and uh we have uh an IDE-like display that is enabled via Python um and a couple of popular graphics rendering packages that should be widely available for folks so you can still see your code and see the output uh preview before you ever have to do anything with it obviously it requires some facility with Python but that's the trade-off that we've made to kind of enable this sort of functionality um we're not a uh a GUI development uh lab um okay a couple of examples of how we're using this thing we are using this in a liquid printing application I alluded to this um these files can be represented via STLs and the very first times that I I made files uh like this I made them as STLs but that becomes pretty cumbersome pretty quickly um and so this is leveraging the the liquid printing capability that Srasys has been talking a little bit about uh as part of their research package uh this is something that I developed um as a user several years ago uh you can make microfluidic circuits so sort of classic uh mixing circuits you can make some more elaborate mixing circuits uh some fairly elaborate capillary like structures droplet injectors etc this is the subject of a 2022 added manufacturing paper uh that talked all about this liquid printing and and why it works and how it works. So if you're interested in that feel free to reach out to me or check out the paper. As I mentioned we are looking at uh presurgical planning models for 3D printing stuff that you can do with polymers. The idea being to give surgeons and patients more information about what the morphology that's inside them is before these operations occur. You know current models because they're using these boundary representations have a beautiful sort of sparsity but in reality your body is much more complicated than some distilled model. And so we can enable that complexity to be fabricated directly using a volumetric representation workflow. So we take in scan data, CTMR, and then pass it through OpenVCAD you can do the usual sort of filtering thresholding operations, etc. And then directly export to uh PNG stacks if you like. We've also recently explored um mixing materials together in that lattice that I alluded to and showed earlier was one implementation of this. By mixing together three base materials, in this case the tango material, this I think this predated Agulus if I remember correctly one of the Viro materials, the rigid material and then the liquid material, you can get composites which have very large variations in their mechanical properties. So in this case I'm showing elastic modulus, Young's modulus. And so you can get materials that are sub 100 kilopascal um all the way up to things that are gigapascal without switching materials. And that's significant I think because bodies have a wide range of mechanical properties in them. And I'm interested in trying to fabricate models that can recapitulate some of the mechanical properties of the bodies that they are scanned from. So that's kind of a preview of some of the work that we're doing there. I'll skip past that we're doing some print electronics as well. So anyway if you're interested in VCAD um this is one way of getting access to it matterassembly.org is my lab website. So at this point I think that the Google search now understands that when you say open VCAD you want open VCAD so if you just type open VCAD it should take you there as well. We've got a GitHub associated with that that has some of the documentation and information about how to use it in Python. So I also want to just acknowledge my fantastic PhD student uh Charles Waite who's a CS PhD student who is going to be um graduating in the next uh you know I would say 12 to 18 months depending on what he wants to do and he's gonna be out looking for the next next best thing.

SPEAKER_01: 38:01

So uh anyway thanks for your time and uh I'll take questions if there are any yes guys if you have any questions please enter in the chat box or QA actually you know what put it in the QA so I can keep track. Okay well how about commercializing your open source?

SPEAKER_07: 38:18

Yeah so um it's a it's a tricky thing to try to have you know open source and and commercialize it right we've tried to to walk that line by making open VCAD available to hobbyists and academics uh in truly open form uh they have to contact us just so so they can tell us who they are um that's the one uh trade-off that we made with my intellectual property office here at Boulder. There is an interest in also commercializing it we have applied for patents that they cover the underlying technology. Of course the code implementation itself is copyrighted and we control the access to that and we've made that the thing that's going to be available for open source for for hobbyists and academics. And so we're we're interested in trying to get people to use this um and uh you know hopefully it's useful for for folks besides just my lab.

SPEAKER_01: 39:06

Okay I am an outsider but it looks very useful to me. Uh and also one of the benefits of commercialization is that you can standardize your product to make it more easy easier for people to use and also tech support, which probably I'm assuming you're you're doing the tech support right now. Just my student Okay so another question that I don't understand is um all these printers with multimaterial capability I'm assuming they're they can with some kind of multimaterial software to run the machine but there is no there's no way to switch machine from one to the other for the designer is that correct?

SPEAKER_07: 39:42

Yeah so that I mean they're pretty much all predicated on using a classical boundary rep uh design interface, so SOLIWWORKS, inventor, what have you and then generating multiple distinct STLs which then represent the different regions that will have different materials associated with them. And uh You know, that that takes you a certain way down the road, but but only so far. Uh for example, if I have two regions that you know intermesh like this, and then I want to make my fingers a lot smaller and more numerous, you have a rapidly uh growing number of triangles that are in your STL files. Those can become quite cumbersome. Um, and ultimately, if the resolution of your blend approaches the resolution of your printer, you just have unwieldy files, both on the CAD side and on the compile output side, the you know, the boundary rep that you're going to pass to your printer, whether it's AMF or 3MF or STL or whatever. So I think there are some um you know boring but technical reasons why those files are not quite the right way to do design and fabrication. Um so that's the kind of justification on on the side of uh printers that can be represented using B reps. I would say other printers like those that take bitmap representations um really kind of fall on their face when it comes to using boundary reps simply because their resolution is so high that the argument I was making about the number of triangles um kind of kind of breaks down. Furthermore, um you need to use some additional set of tools. If you have, say, STLA, STLB, fine, you can define them like this using a uh traditional CAD package. If I want to then blend between the two of them to make the interface um, you know, a gradient, uh, you need some other tool set that allows you to do that. And um, so certainly there are other folks who have been doing really fantastic work that's based on functional grading, exploits blending. Um, I'm not saying that that you can't do it without OpenVCAD, but I would say that OpenVCAD makes it a heck of a lot easier and gives you a lot more control um over how those blends occur, where they are. Um, you know, if you want to have them locally driven, by the way, based on local geometry, that's kind of a nasty thing to do via other means. And um it's fairly straightforward in VCAD.

SPEAKER_01: 42:00

Awesome. And uh okay, we have a question in the QA box. Let's take a look. Uh, how does the VCAD modeling help in the gradient design of the human skin? I don't know. That's very specific. I was a human tissue in general.

SPEAKER_07: 42:13

Yeah.

SPEAKER_01: 42:13

Uh yeah. It's all layers, so yeah, it's correct.

SPEAKER_07: 42:16

Yeah, absolutely. So we we do have a um an unpublished uh bit of work where we actually are trying to recapitulate the mechanical properties of human skin. And so you can imagine that um, you know, the mechanical gradients that say a needle uh would encounter, or even just simple palpation uh that you encounter as you move deeper into whatever model is you've printed, um those those become quite complicated. And so I just think that it's much more straightforward and more powerful to design those gradients using functions that I can rapidly manipulate using code versus using some kind of uh of graphical click and drag interface. Um I think that probably the tool that you would go to if you were trying to stick in a in a GUI environment uh would be to kind of build up your own sort of sketch using Grasshopper and Rhino. Um I've seen folks who have done that. Um hats off to them. Um I think that um uh VCAD is is my tool of choice.

SPEAKER_01: 43:19

So I'm assuming this is a VCAD version number 1.0. Is that correct?

SPEAKER_07: 43:24

Or is this like so if it were if it were commercial uh uh software, I would be more um uh I guess strict about the versioning. Uh we've just been kind of doing the versioning based on what paper is out right now. So this is the Python version, and the Python version is the version that you can get uh with the Python package manager tools.

SPEAKER_01: 43:45

Okay. So I mean, in terms of uh the next thing that you want to do with your current version or whatever iterations, um, what are you looking for, let's say next two years? You know, what are you hoping to see?

SPEAKER_07: 43:58

I will I will allude to the next paid publication which should be coming out this spring. Um and so in general, we're interested in enhancing the ability to specify desired mechanical and material properties as a spatially defined function. Um currently, users of 3D printers have to know a lot about what printer they're using and what material they're using. Uh and then they have to go backward into their design space to design something such that they will achieve those desired mechanical properties. And we're working toward making OpenVCAD another way of handling that.

SPEAKER_01: 44:35

Interesting. Okay, we'll look forward to that. Thank you so much, Professor. Um anybody let me see. There was one more question just answered. Have you tried exporting into AMF format for bending materials and colors in a mesh? Blending, sorry.

SPEAKER_07: 44:50

Blending, yeah.

SPEAKER_01: 44:51

It was like, well, bending is also kind of interesting too, but yeah.

SPEAKER_07: 44:55

So um I I don't think we have implemented AMF. Um and we could, um but we have not yet done that. Yeah. It's a nice suggestion.

SPEAKER_01: 45:08

Yeah, well, Bob, if you pay, you know, you could probably get that earlier. I'm just kidding. I'm just joking. That's a joke. Um, all right. Thank you so much, Professor. Uh like to introduce our uh lesson because we have a packed schedule, let's move on to our next speaker. Thank you. I see. Okay. Our next speaker is uh from uh the fantastic RD group from HP 3D printing, Nathan Shirley. Nathan? Yeah.

SPEAKER_06: 45:41

Yeah, um, let me get my presentation up. Okay. No, really interesting to listen to both both previous speakers, um, very relevant to what I've experienced and and hopefully what I'll be talking about here. Um a little bit different format for me. I'm presenting um some work that we've uh reached through collaboration with radii devices. Um Josh isn't on the call, but I'll be representing them in this context. We were introduced to each other, the two groups from from HP and from RADI through collaboration that the VA has been leading. And so we've spent a few years now with uh with these two teams along with a couple others implementing solutions for the VA to digit to digitize their prosthetics um process. Um so we've seen a lot of the benefits of uh some of these collaborations and have really come to the conclusion that um it's better to work together between experts across you know very different fields. Um the premise here is there's a lot of very interesting technology coming out, whether it's for additive or for AI or for what have you. Um but without real applications, it's abstract and pretty uh meaningless in its um in its use in its usefulness for for people. Um and we have found uh for for MJF, for our HP printers, and for our team, that I'll explain in a little bit more in a second, that orthotics and prosthetics is a is a not not only a very relevant application for uh for our additive uh machines, but that it's a very real and rewarding context that gives you uh very um visual feedback if you've been successful or not. Sometimes it is kind of abstract. You don't know if you can accomplish the job, but when you see people um now mobilized um as a result of the work, it's very tangible. Um in terms of prosthetics, I'll get into a bit more here, and this is more Josh's realm, but I'm I'm becoming quite quite familiar from from my viewpoint that uh the fit of a prosthetic is the foundation um before all else, kind of like you would put a foundation under a house and you want it to be sound. Um you it doesn't really matter what else is in the product if it doesn't, if it's not comfortable uh to wear on a daily basis. The other thing here is that there's a there's a pretty large population of people that need access to good prosthetics, and the current clinical context is not scaling or scalable in the way that uh the population needs. And so you need not only automation on both ends on the clinical side for how the interface of the part and the body are designed, you know, with a lot of uh clinical context. But also the product itself is a product like any other, except for it has to be customized every single time. So this concept of design automation takes precedence over kind of like um more static uh design uh processes where you you work for 18 months on a product and you make one design, you make millions of them. Um so Radii is leading the way on the clinical fitting side. That's why they are, you know, the VA reached out to them, even though they're based in the UK. Um and HP is forging the path in this context for design automation. We've basically built an internal team that acts as consultant for uh people who have purchased our printers and have applications that are uh whose automation and design would not be um easy to implement with the the available tool sets. Um I totally see the the advantage of tools like Open VCAD or some of the process, the ones that are more um like Grasshopper, uh like um NTOP, Blender Geometry Nodes, or in our case we leverage Houdini quite heavily, where you can really program uh what you want to happen in a design context. I alluded to the population of um people that have uh the need for sockets and 170 million people worldwide, um, and every single one of these is made custom by an expert called a prosthetist, and the availability of those experts varies dramatically throughout different countries and locales, which uh so some people are getting better care than others if they're getting any care at all. Um but one of the main things is that a poor device fit will cause pain, injury, and you know, decreased mob mobility, um, and requiring months to years of refitting processes with uh you know unsatisfactory outcomes, leading to about 31% of all prosthetics made are just abandoned by the users. And they and that puts them back into a wheelchair context or or spark crutches or something that's gonna make you know things uh not ideal. Um the traditional process for creating a prosthetic socket is extremely bespoke and artisanal and manual, but with a clinical eye. Uh it's done through casting and um and then uh you rectify the shape, which I'll get more into so that it's comfortable to wear. There are multiple test sockets usually going through the process to dial in that fit where they can they do a clear socket they can kind of look at and see the interaction with the skin and do uh walk tests. And then they make the final definitive product um that is hopefully wearable for uh for years potentially, uh depending on the patient. And then um, yeah, and then finally it's it's fit and they they take it home. Um but it's kind of hard to do, and I and this is of course in radii's domain, um, but I'll so I'll speak a little bit on their behalf. But if you just take the scan um in a digital context of a person's residual limb and you load that into a device of the same shape, uh it's extremely painful to wear. It would be unbearable because you're loading the the bone um where it's been severed and the scar tissue, and you would probably lead to further amputation if you uh if you simply did that, if they were even able to wear it. Whereas what you really need to get to is more of like a press fit from the sides, and you need to understand where the where the hard the hard and soft tissue are and and and kind of suspend them from that so that the there's no real load bearing at the at the base of the prosthetic. Um here's a you know image of kind of like some of the the ideal loading zones and the keep out zones, if you know, in terms of pressure that you have to accommodate. And so that leads to some pretty complex uh design intents uh from clinicians. The two prevalent ones being uh patella tendon bearing design, where you can see that there's that red line right below the kneecap where a lot of load is able to be borne, along with kind of around the calf um areas. Uh or there's a total surface bearing design, which kind of takes more of a universal global offset approach around the um cylindrical form to try to kind of hold it up. And these are um, you know, through prosthetists and um will really go back and forth on which one is best. And then they all say they have their own kind of secret version that they've come up with over time. Um so the challenges that are there is that um we're kind of nearing the limit of innovation you can do with traditional fiber, uh carbon fiber uh fabrication. It it's a very um impressive material, but ultimately you have to hold it together with resin and to incorporate the other elements that come in. It doesn't take full advantage of its strength when you have to make these interfaces. Um adequate fit is a challenge, uh, lots of abandonment. Um and the this the workforce, whether they're the actual prosthetists or the uh the clinicians or the ones more doing the the fabrication, which might be more of a skilled laborer context, they're very they're not readily available and they have to be specially trained. Um there's a there's a dearth of such and uh people. Um and then in terms of their resources, um the there's places where there's just a kind of a vacuum where there's not availability, whether it's low-income countries, conflict zones, people uh with higher, you know, more amputations are occurring and they need to get access quickly and at scale. Um and there's really kind of uh because everything's done analog, there's not really a great record-keeping set of you know what works. And so um that can lead to very um different levels of care um throughout the the world. So uh what RADI has done initially, and they've expanded upon this, is they they took a set of uh sockets kind of um of the original LIM and the socket that was designed for that, and also measurement scores on outcomes, and they've created a um you know a model that can pull out all the different modes of design um uh and create a recommendation, but also one that's highly controllable in an intuitive fashion for the clinician to then go and do their work. So they follow the 8020 model where you're giving a representation, uh a recommendation, the kind of 80% is a starting point, and they're you're expecting them to then leverage their clinical expertise, their knowledge of the individual patient having felt the limb, and then dial it in and spend more time on the on the finesse than just getting to you know the starting point of every socket potentially. So one thing that they found through this through this data collection is that you know, while you had telatone bearing and total surface bearing, and everyone says it's kind of one or the other, that the data actually suggests that everyone's doing a hybrid to some extent, and that that's also what's most successful in terms of comfort. Um so then they, you know, of course, there was a process of figuring out what um, you know, is it successful? And you know, there's not necessarily at this point um to get to something that is dramatically better, but they're trying to prove that they are uh on par and and not and not any worse than what you would get. And so they've done some some work to prove that they are you know actually um appreciated by wearers and that there's a and some of them have claimed that that's their their favorite socket that they've received. Um because you're incorporating all the knowledge of all the prosthetists that were involved in creating this, you know, in this model, basically. I'm gonna skip through some of this, but basically they're they're in a good stance. Here's a little overview of how the interface would work for reform, um, where you can uh you get these recommendations, but then they're you know intuitively controllable in terms of the offsets that you might do uh to get that um truly dialed in. I mean, I have another video that shows a bit more of this, so I'm gonna just kind of skip from here so you can kind of see that the clinician has total control of what um elements and offloads and loading areas that they want to do. And then once you make all those edits, you can still ask the model to massage them so that they're working in concert in a way that the model suggests, so they'll kind of you know do a little bit of um work at the end to kind of give you the right overall effect based on your input. So they've been having great success um in terms of improving comfort um uh through through this route and good attention in the industry as well, kind of leading with a data first mindset. Uh so I'm gonna, in the interest of time, I'm going to kind of make the transition now to I've been talking about this kind of interface with the human body, which is key. It's the foundation of everything. But at the end of the day, there's still a product to be designed. Um, and there's, you know, in the the carbon fiber context, it's uh very um there's a lot of known ways to do this, and it's a very labor-intensive process. But in a 3D printing context, um you have a lot more design freedom and a lot more ways that you could also mess up if you're just going from the first time for the first time through in a digital context. You don't really know what equivalencies there are, how to how thick to make things, how to transition, um, what kind of uh aspects make for a better connection to the distal geometry or hardware that goes goes on to the socket. And the software landscape is incredibly dense and confusing for people coming from a clinical context. There are some that just are passionate about it and they dig in depending as um as Matt kind of alluded to on their budget. There may be different ways to go. But just from picking the scanner, how do I then actually do the rectifications? Radi is on that list right there. And then if I get into a CAD context, most of these fall in their faces if you're trying to get into a BREP NERBS context, like Rob said, it's just not really um helpful to go backwards to BREPs when you're starting with mesh data and you're going to print mesh data potentially. Voxels become much more useful because they um they just don't break like like um mathematical CAD does um in terms of the way that surfaces are handled. Ultimately, our team is completely focused on this design automation standpoint. We can do we can create all the geometries that you do in traditional CAD, but in a way that is completely robust for new incoming patients and adaptable. Um, so here's uh an implementation. It shows um the reform software from Radii, bringing in a new scan, getting the recommendation of fit. You know, they do some cleanup uh of the scan, they get All their adjustments in to kind of understand the axis that's important for the geometry that goes on the bottom. And then what you'll see at the end of this is that our team built a design automation that is basically a one to a one-button-click socket. After you've done all this, you just say make socket, and you don't really have to worry about any of that. There might be in the future some choice points about what kind of connection you want at the bottom, maybe some variables to play with. But essentially, we wrote a script leveraging Houdini that in the back end on a server just computes a custom socket based on the input surface that Radii outputs. So we kind of pick up the baton, run a few laps in the background, and then deliver it back to the clinician to where they could then order it from one of our service bureaus or if they happen to have printers themselves access to them if they're internal to a company like that. So here the you know, all of those rectifications are happening, all the important stuff of a good socket fit, and and then it basically once you have the trim lines and everything, it's just send it for processing and out the other end, you get this final socket. This is a very simple 3D printed socket, but it has all of the hardware and all of the considerations to be robust. Here I'm we're showing kind of like a what could be uh in the future as we develop you know more of the uh potential of the 3D printed context. This was more of a show car piece, um, just to say to get um clinicians thinking about what's possible. Um and then with uh I guess just keeping track of time here, Jenny, do I do I have a few minutes to go through the process on this one? Or are we kind of thinking?

SPEAKER_03: 1:01:51

Go ahead. Yes.

SPEAKER_06: 1:01:52

Okay. Um so this is just kind of like how I'm an industrial designer. I'm trained traditionally in CAD and have designed products um in the housewares and in the um electronics context um at you know at various companies. And there's a whole different way of thinking to get to a similar result that's much more adaptable. So in this case, this is the scan coming in from Reform into our Houdini context. I'm showing the same scan from two angles so you can kind of keep track of what's happening without me having to rotate everything. But you know, the in it the first thing we want to do is maybe put in a little bit of a flare so it's more comfortable around the perimeter. And then uh and then we're gonna create kind of a subtraction tool so we can always remove the limb from what we design moving forwards. So here's here's kind of a quick way to do that that uh removal of the interface with the body once we have the that. And then moving over, we've taken a scan of the sound limb uh in this case, and are you know, you need to kind of incorporate that for this this design, which is the show, the kind of future context I was referring to. But when you put these two things together, they're incongruent. They you don't really have, you know, if you got those two bodies in CAD, you'd you'd have to do a ton of custom surface patching to figure out how to kind of leverage both those scans after you converted them to the right context. But in a mesh context, we can do voxel blending and radii applications, so minium minimum radius on that, do some smoothing processes. And then what was very useful in this context is we used uh attribute painting from that perimeter to say I want to respect the sound limb reference scan from the other leg in the black areas, and I want to respect the edge of the incoming um fit from radii in the blue areas, so that it kind of does a transition based on um on distance. And that way now you see it's kind of tucking in right as it comes to that line, so you can get a nice thin perimeter when you do the actual um subtraction. So now, so now we've incorporated the exterior and the interior and made sure that at the top they they stitch together nicely because of these fall-offs. Um so now you can see from a cross section it is variable thickness, but that top perimeter is consistent. Um to get out of the perforations, we're doing things like point scattering to create some um some Boolean tools for that. Here we're kind of getting some of the details of the design back in by then creating another point scatter and doing a larger Vornoi off of that. Um we could, there's all kinds of you know, better lattices you could leverage where I can see VCAD. If I had access to that as a Python context inside of Houdini, I would be leveraging some of its capabilities because we have to write them ourselves, and that's kind of redundant if someone's already working on it. Um so ultimately, you know, these pieces come together, and then the final surface is shaved away, so you're left with the design of the device itself. And of course, we would need some distal geometry attachments on the bottom, and now the perforations have been removed, and uh you know, then you have this thing, which would be uh basically impossible to model in a traditional CAD context, but requires a computational tool. Um and uh so I'm just gonna let this play in the background, maybe while we're talking. This is a customer we have that did did this uh internally on their side, but they are, you know, they they are still spending um you know four to eight hours to design digitally. And so, you know, companies like this can benefit from having design automation where this entire design would um would be able to be output automatically, you know, from an input scan and some some choices from a clinician. So I think to to comments that have made previously, the hardware in the additive space is incredibly capable at this point, but the software um still requires um a lot of of work to get to a point where it's leveraging this the hardware appropriately. Uh I'm gonna stop it there and take any questions.

SPEAKER_01: 1:06:03

Um Thank you, Nathan. That was a very good presentation. Um but due to time reasons, I think I'm gonna move on to the last speaker before we come back to you. And also, I'm not saying too many questions in the chat box, so I'm gonna wait for people to marinate on what you talked about, which is very packed information, I have to say, but very useful. Um so next speaker, or last but not the least, is um Alexander Gint. And he has a company test a seat, which has uh have fascinated me for a couple of years. Hopefully, you know, eventually I'll see them more. So, Alex, take it away, please.

SPEAKER_00: 1:06:41

Thank you. I think it's also great timing to kind of more explain about the product, and it also speaks about a lot of those issues and concepts that uh previous presenters also were speaking about. You can see the presentation, right?

SPEAKER_01: 1:06:56

Yes.

SPEAKER_00: 1:06:57

Okay. So uh my name is Alex, I'm industrial designer in my background. And about 12 years ago, I started a company called TestaSeed by working with the rehabilitation centers and hospitals and working specifically with families that have a child with cerebral palsy. And what this means is that those children burn with uh disability, that they're not able to sit independently. And we're usually not thinking about sitting so much, but sitting is really a center of everything that usually children are doing if it's from education, eating, meals, playing, etc. And that means that those children really depend on the sitting because they're not able to sit independently. I think in some way it's really similar to the prosthetics field and the orthotics. So it's really crucial. We usually think about prosthetics as something that's really crucial for a person to be mobile, but sitting for those children is almost the same. So just to explain a little bit more about the sitting, let's dive for a minute for a daily routine of those families. So a child with disability that needs a sitting solution, they will start from uh morning routines, they need a sitting solution for bathing. Uh, they also need something to eat and maybe be more mobile in their houses. All of those uh equipment's extremely expensive. But also you can see that they are far away from being any customized, so they're not really supporting well most of the most of the children's. But then also, in most of the cases, if you want to be more mobile, you will need a car seat that it's a separate product. And if you need any device in your school, you probably will need an additional one because you can bring your own from your house because it's too bulky. If you are a traveling person and you want to travel with your family and your kid, you need more portable device, etc.

SPEAKER_08: 1:09:11

How's their own morning? Uh how's their own morning going? About to have a mental breakdown, how about you? Nice, nice. I'm in the middle of a mental breakdown already.

SPEAKER_00: 1:09:20

So, for those families, like sitting solutions are so crucial, but they are also so complicated and expensive. So, in the pediatric space, you can imagine that this is pretty much a banded space because no one wants to make custom sitting solutions, especially for pediatric space, because they're growing. So, in the end, most of the families, if they're not able to get all of those equipment, and also every three, four years they will need new equipment because they grow outgrowing them. You will find that most of those families they have those DIY products that they make them from pool noodles, uh, cardboards, etc. etc. So even if they can get adaptive equipment, it will not fit them well because it's designed really in the way to fit most of the people without customizing them. The biggest problem with seating is that the child in the early age, if they're not supported enough, they start to develop spinal deformation, they're not able to participate in the education or even uh in kind of playing with peers, they will also start to develop uh development delay and more and more and more that will cause them a lot of different treatments in the future. So it's really crucial. The problem with seating is almost the same as the prosthetics. To make a custom seat, it's a it's a it takes about like three to six months to make the custom seat from scanning, manufacturing, fitting, refitting, testing, etc. So it's not really an option in the pediatric space, even though it's really important and crucial in the early stage. So this is why we started TestaCed. And TestaCed is really trying to focus on three different aspects. It's one, really make a custom and more fitted sitting solution for children to provide them better support. But also we're trying to see how families are actually using them because like customizing them and make the child sit, it's one, but really how the families use them, and then also make it more affordable than any other products, and that's all really possible by 3D printing. So you can see that this product is the original product, and those therapists that we worked with them, they like the product, it's super comfortable, they it's super supportive, but it was made like 30 years ago by some company that was customizing them by hand, and now no one wants to make them. And this was a starting point for us, thinking um it's a great product, therapists like them, families like them, but no one manufacturing them, let's digitalize it. So we started to work with the therapists and the professionals in this field of sitting and positioning and see what kind of measurements and what kind of supports and what parameters we need to include in this model. And we slowly started to develop the product into the digital space and also configuring it in a way that we can take measurements and then also digitalize the product without any scan and uh after processing of the product because for us it's also a thing that allows us to scale because scanning and then processing is one of those gaps and then prototyping. So 3D printing really allows us to start as a small business from prototyping, providing some um basic products, and then huge learning curve of what works, how works, uh, in which way we need to print. But it and and and and also validating, and this is, I think at this moment it's less bothering the customer's therapist. But 10 years ago it was one of those issues that people even not trusted 3D printing. Now people use 3D printing products without even noticing, but it's a process that we also validated the 3D printing approach that we had using FDM or XGF, that we actually can create uh pretty solid and rigid and durable material by this approach. Now, the end result of Testa seat, it's the more updated product you can see. It's also started without any attachments, headdress, footrest, etc. But it's really between a seat and a brace. And its goal is to provide a best support for a child. So we integrated all the side support, hip support, pomels, headdress integrated to the seat. So families usually that buying the equipment, they don't need to buy the equipment and then all the attachments. And the way that we figure it out, it's really creating a customization platform. And it's for two different reasons. One, we know that there is no one sitting solution for all kids, but also there is no one sitting approach for all the therapists that we're working with, because some of the therapists will tell, hey, we want the child more tilted, or we will want more straps, etc. So we really quickly found that it's kind of a collaboration between us, the therapist, the family, and the kiddo, and we're trying to figure out what measurements and uh what approach the child will get best. And based on the measurements and the therapist feedback, we really can enter the measurements without any scanning, and then based on those measurements, we can create the fully custom seating solution. Now it's really complicated to explain how customized it can be, but really almost every single parameter of the seat, if it's depth, height, side support, hip support, um, tilt, recline, can be adjusted and customized. Also, recently we started kind of doing more diving deep into the components of the seats. So the footrest, the seat angles, heights, different variations are also customizable. And we're working with the therapist to kind of define what is the next custom seat or what is the uh next attachments we need to work on. But you can see that the variety of the seats is huge, even though that all those seats are in between the ages of one to eight, they're really different, and for some kiddos, they're not able even to sit without having this kind of custom seating solution. Now, we're having two major benefits from using 3D printing and the approach one, yeah, we can position them much better than any other uh product in the market. But also you can see the footprint, and this is something that we're discovering by really working with families. The small footprint that we can provide and the lightweight of the product, it's a key for all the families. So we discovered that yes, customizing is a basic because they need to fit, but then the the the low profile and the lightweight is the second priority for all the families because they need to take the equipment with um with them for traveling, schools, families, etc. And then the other approach is also how we can create the product that can be usable in different setups, like sitting on the floor, sitting on the chair as a booster. And it's also something that by 3D printing, we always uh redesigning and improving the product. And just for example, how families can use it. We'll just dive into our social media so you can see feeding and meals is so crucial for those families. And then by having the opportunity to take the equipment with them, they actually can take go to the restaurants, events, they also can sit on the floor so they can start more uh interactions with peers and friends, and getting more of the educational piece. It's also water resistant, so we can we chose materials that and the shape of the seat that it can be easily clean and water resistant, so families can use them in shallow water, but also as a showering chair for some of them. And the small footprint is something that's always kind of exciting to see, and we discover keep discovering how families actually can use it. So bicycle trails, um, they use them also as an insert for strollers because most of the strollers are not supportive enough. It's so small that it can fit in different swings, they can use them on airlines, and we are so excited that sometimes those families are discovering new opportunities that have never been possible because of the size of other products. So, like sleds and mobility, it's one of those things. But we're also discovering that not only families benefit from them, we're also seeing a lot of therapists start using them as an equipment in their clinic. Because previously it's also too bulky for most of the uh therapies, and it's also not something that they can use as a in-between brace and the seat. And really, the ideology that we have is that we want to start with the child as soon as possible, provide him the best fitting solution, and then grow with him if it's the size product that we have or the custom ones. We have both of them. But we have already children that are growing with us for like eight years and benefiting from the support, but also the portability. We're kind of a weird uh hybrid company in this space because most of the companies you have size products or a custom fit, and the custom fit it's not really exist in the pediatric space, so we're kind of combining both of them, and we're also providing a rental program. So it's also something that because we're manufacturing, we feel pretty well to kind of make the rental program as well to make it more affordable and available. And we still kind of navigating ourselves into the reimbursement system because it's not easy in this space, but we was able. Able to deliver already about seven, eight hundred of the seats to clinics, families in US, Canada, Australia, the 3D is seeing huge benefit and difference between other products. And again, it's all thanks for the 3D printing and the different properties that we can make, but also the learning curve, it's long. So we're also adapting the products on the way. And really for us, we know that now we're focusing on the pediatric seats because there is a huge gap in the market. But we also understand that the same technology, we can use it not only for pediatric space, but also wheelchair space that starts to be really also looking for those custom solutions. And we already tried out some of them. And it looks like a great, great opportunity for this market to grow into the more customizable solutions than off the shelf. We are self-sustainable right now. We are cash positive, but we are looking for investment to grow. So if there is someone in the crowd that is looking forward to invest or kind of join the efforts that we have, please share. Please reach out. And yeah, if you have any questions, would love to share more.

SPEAKER_01: 1:21:22

Thank you so much, Alex. I've seen uh quite a few iterations of your products over the years. Uh, it always fascinates me, and some somehow my hope of getting my own personalized seat is uh disappearing. Uh so how did you decide of uh which population to design for? It seems like you have we have a lot of problems with our seats, but how did you narrow it down to the pediatric disabled children population at the end?

SPEAKER_00: 1:21:51

So I believe that in the future we will also make them for an adult. I have challenges to have my seat, my office seat. I struggle. We are sitting so long on those seats and they're not, but we're not really choosing them. We see that they have the biggest needs. So children with cerebral palsy, brain injury, spinal injuries, uh, spinal befita, they just are not able to sit independently. That's all. Like without the seat, they are falling. So they are kind of the first market that we see that they need the seeds, and there is no good solution on the market. So this is why we choose them. But then growing in the future to the wheelchair market, that yeah, they they have a lot of more advanced products, but still the price point, the customization process, the weight, it's much more higher than we can provide. And then once we figure out with all the disability space, we actually want also to get to the office chairs.

SPEAKER_01: 1:22:58

But eventually.

SPEAKER_00: 1:23:00

It's it's it's a different market at all. So we want to solve the challenging pro problems and then hopefully getting also to the to the office chairs and other products.

SPEAKER_01: 1:23:11

Awesome. Okay, so I would invite all the uh panelists to join. Um, I'm not really sure. Let me see. Oh, yeah, we have some questions in the QA. Let's see. Um questions for Alex uh from Parash. Uh, do you know if such solutions has any safety or regulatory guidelines? Yeah. Talk about the regulatory process if you can.

SPEAKER_00: 1:23:33

Yep. So we have two different regulatory one is like general floor activity chairs and boosters that you we need to be compliant with them. And then on the FDA side, we are still trying to figure out what is the regulations that they require requiring from us, especially on the customization side. Because we we're not just capturing the shape of the body, but also like adding a tilt or a client to the seats change the entire dynamics. So we're trying to see, and we're still kind of learning in this process if someone knows a good advisor in this space, how to get the product really get with the FDA. We are class one exam, so it's pretty easy. But the regulation there on the custom products, they don't like the custom words, especially if it's too customized. But yeah, so it's a general regulatory side from floor activity and booster, and then FDA, we're still figuring out.

SPEAKER_01: 1:24:36

What is your uh experience with European market?

SPEAKER_00: 1:24:40

Um not on the regulatory side, but because we still in really low volumes there, so we have families that are buying them directly. Um from the clinics, once we will clear by FDA or register by FDA, it will be similar, but they struggle with the same issues in this market. Really, families from Australia and Europe market, they are reaching out because in many cases they're just not able to find any seating solution in the market that can fit the child. And this is something that we're seeing a lot from Canada as well and Mexico and other countries.

SPEAKER_01: 1:25:18

So you definitely found a group of people with a huge unmet needs.

SPEAKER_04: 1:25:23

Yep.

SPEAKER_01: 1:25:24

Yeah. Okay. Okay, I'll welcome all the speakers on the on the camera if you can. The remaining speakers, um, uh Professor uh Rob has to leave for a class. Um okay, another question actually uh for Nathan is um how do you account tolerances associated with printing and filament to users? Not sure that's uh that's a very general question. Uh how do you account for tolerances associated with printing and filament? I'm I'm not really sure like if that's uh the question makes sense to the designers. I'm not it's maybe just how do you manage tolerances?

SPEAKER_00: 1:26:09

So I I would say from our perspective, we are now focusing on FGF, so we're using uh pellets, and I think the tolerance there is probably even lower than the filament, but we are also printing in much wider nozzles and wider uh like the printing with is much wider than the standard one, so the tolerance there it's pretty bigger than then like the micro. So for us it's not the biggest challenge.

SPEAKER_02: 1:26:45

Okay. Sounds good. Nathan, do you have any uh do you have a any thought about that?

SPEAKER_06: 1:26:51

Not as uh involved with FDM, but uh I think that from my my side there's tolerances around um kind of thermal warping and things like that, since we have a the whole chamber or the whole build volume is cooling. But on a part-to-part tolerance, we're actually can be better than um like scaled injection molding for like interfaces because we can print them you know together, you know, and then they they'll they'll fit. But then if you want to do that over you know many, many parts, then you have to build back in those tolerances just because they might be coming from different jobs and things like that.

SPEAKER_01: 1:27:28

Yeah, I just want to mention that any HP projects is uh multi-jet fusion, not filament-based. Um, okay, we have a question from Drew uh to Nate. Is HP Houdini workflow going to be commercialized under the Radii brand?

SPEAKER_06: 1:27:46

Uh so we build we're building currently workflows for you know many of the um O and P companies that you would recognize. Um Radii in this case is a software company, which is a little bit different, and that's why I can talk about what we're doing uh in that case. We definitely um there's a potential for something like that to be commercialized through a provider. Um, I think we're just kind of looking for the right fit um on that side. I I would say kind of open-ended. We want to get our goal is kind of to democratize the ability to do this without having to be a developer. Um, and so the more we can get our solutions out to people, um the better. So Radii would be one avenue, but maybe not the only one, depending on, you know, whether, you know, whether you're a more of an individual or part of a company that's working with us directly.

SPEAKER_01: 1:28:43

Awesome. You know, one thing I learned from this panel is that uh yes, learning the software is important and knowing all the latest technology is also important, but also like be very close to the clinical side of things is super important. It's perhaps harder than the actual maneuver of softwares. I just I just want to hear your thoughts on your perspective in terms of your particular verticals. Like where do you think the key challenges are and what do people need to understand before they tackle that problem? For everyone or for for each, just your vertical. Yeah, so everybody. This is a question is for the entire panel. So everybody has can say you're a part.

SPEAKER_06: 1:29:31

Um yeah, so I think the challenges for for us um are around scale. We're we're you know, as a company, we are looking for applications that will, you know, come completely change the the business structure moving forwards. We had a point where a company that we that we enabled through our technology was worth more than HP at one in one point, direct smile direct on the on the market. Um and that kind of opened our eyes to how important the applications really are. Um so so doing that, but then we need we're always we wanna we wanna help all of these bespoke situations, but really we're looking for how can we get that out to you know a mass impact um in terms of scale. So so that that's where all the solutions kind of point towards what's programmable, what can be, what could be run headless in the background. You know, we've we're we're taking whole departments and putting them onto a server, basically, that runs. And it's an interesting, an interesting thing, and getting all of that input and feedback and those choice points incorporated and allowing people still to interact with it.

SPEAKER_01: 1:30:41

Yeah, very interesting. I never thought about yeah, scalability. Yeah, that's very important for the you know, from your perspective as a large company. Alex?

SPEAKER_00: 1:30:54

I think one of the challenges that we're seeing is actually getting to the customers, or maybe in our case, it's getting through the insurance, or even kind of explaining them what are the benefits, because usually we they will not change any products if it's not a tremendous change for their system financially, but also improving the outcomes of the patients. So, and and and those systems are super complicated to just even presenting the product. So we in some stage just realize that we will reach out directly to the families and the therapists, and especially the small clinics, that they might be much more uh open to new solutions. And just after a year and a half working intensively in the US market, we're starting to get much more attention from the DMEs, the durable medical equipment and the insurance, because now the family is reaching out to them and they are asking for this product, then other products. So it's also something like product is one, design is two, uh, manufacturing is three, but really the how to get to the market is one of those biggest challenges, and really explaining what are the differences we can create and how we can change this market in this space. But I think we're getting there and the market also much more receptive for the 3D printing technology as well.

SPEAKER_02: 1:32:19

Yeah. Matthew, do you have any thought about those?

SPEAKER_05: 1:32:24

So so something I think uh everybody deals with from a regulated regulated standpoint is um a lot of my clients have uh regulatory challenges, which is basically like, hey, we've got this new technology software or bespoke products that are matched to patients. Um but the regulatory bodies are typically lagging when it comes to, you know, sort of it's not that they don't want to enable new technologies. In fact, the FDA, at least for 3D printing, has a great team over there that like they want people to be clearing these products. The problem is that there's just such a disconnect, in my opinion, between what sh the standard of care could be now that we have these tools and where it's at. In other words, like device stiffness. Yeah. And all of these customers, like AA Illuminate is is a perfect example of this. Yeah. Um, the screw that we worked so hard to design has a lower stiffness. The problem is the FDA wants us to test it to existing products and and be as strong as. And I'm like, well, that's not gonna happen because we designed it to be less stiff because that's what the body needs. But in order to convince the FDA that you're going to use a less strong device, you have to bring a whole lot of other data, right? And, you know, even clinical data of like, you know, this this is just clinically not what humans need. Like it's just not. But we've been doing this because we were using technology that's 50, 60, 70 years old and testing it to all these old, really stiff products. So I think I think it's just a mindset, and certainly from AM in particular, I think AM sort of exploded onto the scene commercially, but from an investor and like a return on investment standpoint, it's been really crummy. You've seen a lot of 3D printing companies, um, you know, desktop metal, Velo 3D, like that have really promising, amazing technologies just like go from their all-time high to like 1% of their value. So I just I feel like a lot of these challenges combined with an overhyped nature of AM have have led to a big struggle on like what is this technology even for? I mean, the three people on the call, right, and you two um understand what the value is, but I think that just for a regular Joe Blow who's designing regular products, like he's like, what's the value proposition here? I can just keep putting metal on patients and I'm making money, that the hospitals are still paying for these parts. So it's like what pushes the needle forward unless it's not companies like Alyssa and Illuminate that are like, I mean, going toe-to-toe to FDA with an appeal, of which less than 10% of companies win, mind you, and it was a lot of extra cost and effort. But I I think unless companies were willing to do that, then it's just AM's going to be another tool in the toolkit that that gets used when somebody wants something flashy, but not when they want something that that's gonna make a difference. I don't know. Just I'm a little jaded at this point. I'm trying to I'm coming back from my jadedness, but don't be jaded.

SPEAKER_01: 1:35:21

Don't be jaded. We're all in this together, first of all. And two is we're in the healthcare space. It is a complicated problem. I mean, even Warren Buffett, Microsoft, and whatever giants wanted to reinvent healthcare failed. So, you know, what what you guys mentioned really are no challenges for design. I mean, you you can make a lot of things happen. I mean, really, it comes down to have to having to adapt designs to the existing healthcare system is is is what it looks like. And also the economics behind things. And I mean, the US healthcare, you know, we all know it's not really working out. And uh, you know, part of the problem is the misalignment of incentives. And incentives is a huge thing behind whatever that you want to happen, either the uh thriving AM industry or a particular product's taking off in the clinic because of the reimbursement and payment structures. And families who really, really want it. I mean, everybody has a different incentives behind their behaviors. And to manage all that complexity, you know, it it looks, you know, it seems like the design has to adapt to those different kinds of complex environment. But you know, if it is not a hard problem, it's not really worth solving. And uh okay, we cut we have a couple questions in the audience. Okay. Okay. Uh does okay, Drew says does HP has a plan for addressing the orthotics market and design challenges a similar fashion to the prosthetics workflowed. The orthotic brace AFO market. I don't know.

SPEAKER_06: 1:37:05

I can speak to that.

SPEAKER_01: 1:37:06

Um Okay.

SPEAKER_06: 1:37:07

So there's a number of companies that we we collaborate with or that use our our printers that are enabling design for those contexts. Um I can, you know, there that that there's solutions out there um that you can integrate with. As far as like our team, we we work, we found that it's most effective to work for hire. Um so we are um we're we're developing solutions for specific businesses that have a need. Um so as far as you know, if that would be available to all these other contexts, yes, but more typically through those avenues of you know, some of the bigger players that are already out there, unless somebody has, you know, a project that they want to engage with. Um, we we look at those all the time to enable. Um, but we're very kind of open door in terms of also just kind of getting people up to speed. So I I have conversations with people who are like, I just want to get started in Grasshopper or Houdini, what do I do? We just you know hop on a call, I show them the ropes. Um because we tend to our intent is to train groups to do what we do, not to sell them a software um per se. So so that's kind of how we disseminate that ability. Um this question specifically mentioned like AFOs and braces. HP built a whole um orthotics and context and sold it. So our intention is not to create the businesses themselves uh or the solutions themselves, but to enable the players uh in the space and the innovators in the space to do so with their expertise. So we're kind of more of a support function for for innovation, so they can kind of focus on what they're good at and get the help to do the automation and design part.

SPEAKER_01: 1:38:59

Well, the key thing is everybody's gonna use HP printer to make those parts.

SPEAKER_06: 1:39:03

That's help. Yeah.

SPEAKER_01: 1:39:05

And what was the name of that company? Because I I remember they spoke a couple times here. I forgot I'm blinking at the end. Yeah, Rise, yeah.

SPEAKER_06: 1:39:12

Okay, cool.

SPEAKER_01: 1:39:13

I mean, related to that question is like uh if if if someone, if people uh you know, the speakers can offer any kind of learning channels or educational platform. Unfortunately, we don't provide that other than this kind of webinar. Um, do you have any source for people resources?

SPEAKER_06: 1:39:31

I mean, that there's a number of places I a lot of people get started in Grasshopper and then move into some of the other contexts, and there's um a really good, if there's more of a design context, Academy XYZ is a really good one. Okay it's paid, but that's where I usually point people to start off for some of that. Um because that wasn't around when I started. I wish it had been. It was a it was a bumpy road to get into computational design.

SPEAKER_03: 1:39:56

Yeah.

SPEAKER_06: 1:39:56

Um, and there's other ones as well, but that one seems more productive. Focus than a lot of the other ones are architectural focused and so they're kind of tangential.

SPEAKER_03: 1:40:04

Any YouTube channel uh that people can follow stuff like that?

SPEAKER_06: 1:40:09

I mean the kinds of YouTube channels I'm checking out are probably a little like um Judy what's his name? Judy Chiro um does some really deep stuff in Houdini, but that's a um that's a different context when you get that that high.

SPEAKER_01: 1:40:24

Okay. And do you I mean for people um who want to really get into this, but do you think do you need some kind of degree, like a basic hat experiences, or what's the the you know prerequisite to be able to design for 3D printing?

SPEAKER_06: 1:40:41

Um I would just put out there that our team is comprised of designers, engineers, computer scientists, and firmware people. So it's not a specific one, but you do need to bring some kind of expertise to the context of computational design and design for metal and applet. And the nobody kind of the ability to play in one sandbox is really what the value is. We can kind of speak our own expertise in the same language and collaborate inside of that that uh open environment.

SPEAKER_01: 1:41:10

So Alex, yeah. Alex and Matt, do you have any offering in terms of educational resources for people who want to expand in the into the area?

SPEAKER_00: 1:41:20

I think the biggest challenge today is that 3D printing is so wide. Like FDM and FGF and and all the other technologies, they they they they require so different toolkits. So I think choose the technology, like we focusing on FDM, FGF. There is so many different approaches, how to design materials, technology, etc. And then all the powder-based and and other day they're requiring more of the coding and like it's different environments. And yeah, there is tons of different resources to learn, but yeah, choose the niche, choose the technology, get one, get access, and start just practicing. I think in these days, especially with AI, if you know what you need or you want to learn, you will find it. It's not a it's not a challenge.

SPEAKER_01: 1:42:11

Yeah, that's a really good point. And also we have new 3D printing processing invented almost like every month. I see something new. And it is true the software side is not really catching up to the hardware development, and people kind of omit that. Um, Matthew, do you have any thought on where people can get some help to learn?

SPEAKER_05: 1:42:32

Yeah, so um I'm I'm friends with uh a guy named Duan Scott. I can put his info in the chat. So he runs a CDSAM conference computational design for AM. It's obviously kind of catered towards a more professional market.

SPEAKER_02: 1:42:48

Yeah.

SPEAKER_05: 1:42:48

But he always sort of brings in on highlights really cutting-edge use of computational design, and that has started to include things like machine learning, um, uh sort of as direct applications for design. So if you know, and and this last conference uh um that I was at, the New York one a couple months ago, we had a lot of like more enthusiast type people that were just coming to learn from like some of the leaders in the industry. A lot of stuff on the MJF side, footwear is really big. Um, you know, healthcare is a little bit more, I would say, outside looking in for a conference like that. Other professional conferences, you know, um ICAM's a really good one, international conference on IAM because it's very technically sort of research focused. Um that's probably the two things I would say is at least as far as like from a professional development standpoint. From a software standpoint, man, I just you can't like go wrong with with searching in YouTube and stuff. Like there are like especially for things like Blender or um Houdini, um, even Nentop to a certain extent, although it's that's more commercially facing, but yeah, the ability to just like there's just tutorials for just about everything. But I I agree with Alexander, like you really have to pick a pick a vertical because because it just goes so deep. I mean, like metal additive is where I've sat for a good portion of my career, and the problems you try to solve are vastly different than like an MJF problem. Um maybe, maybe not like when you think of a design for AM in general, there's some similarities for like overhang, blah, blah, blah. But just things like how you deal with heat and feature sizes and crack propagation for metals versus plastics and fatigue stuff. So that like surface finish, all of those things are so very targeted. So I absolutely agree. That's a great piece of advice. It, you know, there, I don't think there's very many jack of all trades in AM. Good, good jack of all trades. Like you really have to focus and say, what do I want to do special? I think Nathan and I have a little easier computational design is a little bit more broadly applied to design problems, but even such, you still have to understand um and have like a domain-specific expert who's telling you, here's what I'm looking for, the problem I'm trying to solve. And then we say, okay, well, computational design can do this for you. It can solve this problem for you.

SPEAKER_01: 1:45:05

So do you think the large language model and whatever that that's happening in the AI space can accelerate the learning process and the speed of adoption?

SPEAKER_05: 1:45:18

I I mean, my short answer is yes, probably. I think it depends on how close you are to the cutting edge. I like to joke that like, you know, LLMs don't do me much good in solving really hard problems because they require somebody has solved the problem and or talked about it before. Yeah.

SPEAKER_01: 1:45:36

Um it's all junks when I start searching for bioprinting and 3D printing stuff.

SPEAKER_05: 1:45:42

For for those people that are like really wanting to understand cutting edge, like get into like current state of research. So at least for like multimaterial hierarchical computational modeling, Marcus Bueller's lab at MIT, they're doing some incredible things, not only on the functional metamaterial side, but on the agentic AI for like creation of new things that have proteins that have never ever been done before because nature, nature is super efficient, so it'll only combine the things that make most sense from a from an efficiency standpoint. But it doesn't mean you can't create unique proteins. And so like uh he spoke at CDFAM, and I was just like, I'm like a fanboy because that's like you know, if you want to look at LLMs and AI for like that actually do enable design-specific new next generation problems, yeah, that's super cool.

SPEAKER_02: 1:46:35

Yeah.

SPEAKER_05: 1:46:36

As far as like enabling education, again, if if somebody's typed it out on the web, LLMs are good, right? They can they they they condense uh that knowledge down to something that's a little bit more usable. So but if you're on the cutting edge, yeah, you you won't get an answer with AI unless you've got a really good AI that's able to hypothesize and solve its own problems, which again, I wouldn't have thought so a couple months ago, but apparently it's it's being done with some success.

SPEAKER_01: 1:47:02

So yeah, and also, you know,$20 per month of subscription is not gonna get you anywhere. You have to pay the$200 one to actually get to their good stuff.

SPEAKER_06: 1:47:11

I keep hitting that wall with uh yeah.

SPEAKER_01: 1:47:14

Yeah.

SPEAKER_06: 1:47:14

It's like buy more credits.

SPEAKER_01: 1:47:16

Yeah.

unknown: 1:47:17

Yeah.

SPEAKER_01: 1:47:17

So what do you uh, Nathan and uh Alex, what do you guys think of the the whatever I would say, you know, we're officially in a bubble for AI, but there is some usefulness to that for sure. How do you see it transform our industry?

SPEAKER_00: 1:47:33

I think it's accelerate things. Um, I think it's still not there to the more like engineering stuff, like for us.

SPEAKER_02: 1:47:42

Yeah.

SPEAKER_00: 1:47:43

Creating one piece, it's one thing, but creating a platform that can really adjust and and and make different products for different sizes and still keep the same priority. Like we have screws and we have some pockets and and and places for different attachments. AI still not engineering, like it's for sculptures, for general uh products, concepts, yes. I think it will be there. But you need a platform, and also the platform needs to be designed around specific product. And it will be easier, it will be faster, but um I don't feel too much of those advantages yet.

SPEAKER_02: 1:48:30

Yeah, neat.

SPEAKER_06: 1:48:31

From my side, I'm heavily leveraging it um to accelerate my work. And I don't not try to get it to solve the core problems. I'm trying to help it have it helped me build the tool pieces that are missing to get to those answers and to create the automations that we're trying to do. So, you know, whereas my expertise is in industrial design, I'm a visual thinker. I've never been very good with coded syntax. I will, you know, I would I I I can describe my intent and I can understand if it's working, I can talk theoretically, but execution in a in a coding context has never been much fun for me. It's not an interest to become proficient in it. So to have, you know, a well, you know, I've worked with a lot of great people in that area, but you also eventually tax their patience, their their expertise, their something, and to have to have that tool at your side that you can just ask the most inane or the most difficult questions to endlessly is useful as long as you have a way to immediately validate it. I think that getting into a loop where you don't know if it's right or wrong is not useful. But if you can if I can say, I need a script for this little component in Houdini Vex, if I plug it in and it doesn't work, I know that was wrong. I go and look at the errors, I come back again. You know, usually it's three or four times uh in a loop, and I've got now that piece that I would have had to bug somebody in a different with a different background to stop what they were doing, to work on my area of X, you know, on my problem. And now I can just keep moving forwards without roadblocks. And I find it incredibly useful. I think it's only gonna get become more so, but it also is important to kind of then stop and ask yourself like, am I getting smarter or dumber in this process of doing this? Like, what am I, what am I really, what's the intent of all this? And so I'm trying to level up in the questions I'm asking and the things I can solve because without kind of just losing the script in the process.

SPEAKER_01: 1:50:36

I I have a podcast. Uh the title is is ChatGTP Making Us Dumb. Uh, you're welcome to listen to that. It's only 10 minutes. Yeah. Yeah. Well, thank you everybody for joining us today. Uh, we're at the end of the year. Thanks for sticking with us. I know uh that we now only have uh a few uh good spirits left, but uh I want to wish you all happy holidays. And uh remember, we're gonna be in San Francisco in person if you are a founder uh or investor. Um the website is uh 3D Hills uh 2026. It's free for um for you guys, most most people who are qualifying. Um so yeah, speakers are all invited for free if you can join. Um okay, well, thank you very much. Uh, this will be on demand for free for a couple weeks. We will also and check check out our YouTube channel as well for highlights and clips. Thank you so much. Okay, bye bye.