Supply Chain Revolution: How Cofactr Enhances Procurement with AI

Discover how Cofactr is revolutionizing the supply chain and enhancing procurement with AI in our latest CTRL+Listen Podcast episode. In this conversation with Cofactr cofounders Matthew Haber and Phillip Gulley, we dive deep into the challenges procurement specialists face today and how Cofactr's integrated data procurement management solutions are setting new standards in the electronic supply chain industry. Learn about the pivotal role of AI in analyzing BOMs, identifying bottlenecks, and ensuring scalability and efficiency in electronic component sourcing. We also explore the evolutionary journey of Cofactr, from its origins to becoming a leader in procurement innovation.

Listen to the Episode:

Watch the Episode:

Episode Highlights:

• Biggest Challenges Facing Procurement Strategists
• How AI Analyzes BOMs
• Will It Become Impossible to NOT Work with AI?
• Is This the End of the Data Entry Specialist? 
• Adapting to Supply Chain Disruptions
• Emergent Technologies that Will Shape the Future

Links and Resources:

• Learn More about Cofactr
• Learn More about Matthew Haber
• Learn More about Phillip Gulley

Transcript:

James: Hi everyone, this is James from the Ctrl+Listen and Podcast brought to you by Octopart. I'm here with my co-host Joseph today. Today we have two guests from the same company. They are Cofactr, you may have heard of them. They are an integrated data procurement management for electronic supply chain company. They've been a few different things over the years, but they're absolutely nailing the space right now. Really happy to have you on the show. Thanks for coming on.

Matthew: Thanks for having us.

Phillip: Pleasure to be here.

James: Do you want to just tell people a little bit about what the company is and what it does?

Matthew: Yeah, so Cofactr works with companies that build electronic hardware, typically complicated electronic hardware and things like aerospace, and medical device manufacturing, but really across a range of industries to help their engineering teams, and their supply chain teams optimize everything from the bill of materials coming out of a tool like Altium, until all those parts show up at the production line at their contract manufacturer in-house. And so, all those steps in between around how do you understand supply chain, manage your inventory data, manage your procurement operations, do that efficiently and in a risk-managed way, essentially. We give them tools and infrastructure to do all that more effectively. We also do the same thing for the contract manufacturers that work with those companies.

James: So, there's a lot going on there.

Matthew: Yeah, we try to take an encompassing approach, I would say, to this problem space.

Joseph: Fantastic. What are some of the biggest challenges facing procurement specialists today?

Matthew: Yeah, I mean, I think before we look before COVID, there was this huge trend towards what we call just-in-time, right? Where you're buying all of your materials the moment you need them, and they're arriving at the production line, you know, days or maybe a couple weeks before they're going to be used, but you're cutting it pretty close. That worked okay most of the time. Obviously, during all the semiconductor shortages we saw during COVID, that did not work at all, and everyone had to switch over to what we'd call sort of just-in-case procurement, meaning stockpiling a lot of stuff up front. But obviously, that creates its own challenges. Maybe you buy parts you didn't actually need, or you didn't really manage your demand plan properly because you have an unexpected uptick in business, or downturn in business that you didn't predict. And cash flow management can be a really big challenge too, right? How do you decide what parts to buy and when? And so, as we sort of come out of COVID, everyone is now having to try to figure out, okay, what's the right balance between these two? And I think a lot of the challenges we see fall somewhere into that kind of decision-making process around, am I picking parts that are safe for my supply chain? Am I choosing to buy the right parts when I need to buy them? Am I waiting too long and endangering the health of my manufacturing? Or am I buying too much upfront and then ending up with cashflow problems? And so, this is a really complicated and rich space, and you need really good understanding of what you own and what you need, and when you're going to need it, and what the distributor base looks like, and all these kinds of inputs to do this effectively. But this is like the big challenge we see across, you know, really everyone we work with these days.

Phillip: Yeah and I would, I would add onto that, that there's also just the time horizon of this and the realities that demand can increase on, on your product. And how do you quickly and succinctly understand as the design that we have today, the bill of materials, is it able to scale? Did the unit economics still work? Are we going to come into unexpected supply chain pinches if we do move from 2,500 to 10,000 units annually? And certainly, all of those sort of problems, both on the what's the here and the now of the supply, what should I be thinking about? How should I be conserving capital? And what does this mean a year, two years, three years down the road? These are all sort of this complicated problem space where historically the data has been really decentralized.

James: Fascinating. So, how does AI analyze BOMs to identify potential bottlenecks, or inefficiencies in the supply chain?

Matthew: Yeah. So, I mean, I think on one level, like a BOM itself is pretty simple, right? Like it's just a list of all the parts you need and the quantities you need them. But then there's all the data that's out there. And one of the things that's awesome about the electronic supply chain space is there's actually a lot of data out there. You know, we look at some other commodity classes like, you know, nuts and bolts or whatever, And there's actually a lot less information out there about where you can get them and who has them, and how many they have. So, one of the great things about electronics is how much the data is available. Obviously, folks like Octopart have been instrumental over the past decade, or so in helping push the industry in that direction. The challenge, though, is the data is, you know, not always entirely correct or entirely complete, right? Distributors are all putting this data out. They don't always have the best handle on their own information. Often they're running on pretty analog, you know, antiquated processes. And so, the question is, like, how do you take all of this huge lake of information that's out there, the information of distributors, the compliance technical information that might be trapped inside of data sheets, or product change notifications. How do you put that together into one cohesive data set that's sort of like true? And then you can actually match that up with the BOM pretty easily and drive a bunch of outcomes. So, AI comes into play when we're trying to look at this kind of big, messy data set, and understand how do you extract the true information out of it? So, you know, really tangible examples are certain types of data are really important to some industries. So, for example, country of origin has become really important for, you know, folks in aerospace and defense, or anyone who's trying to mitigate Section 301 tariffs on their imports. That's not actually super easy to find in a lot of cases. When a reel of chip shows up, it usually says right on the reel where it's from. But until you physically purchased it, often it's not that easy to know that. But that country-of-origin information is out there. It's just trapped inside of kind of annoying to work with materials like product change notifications, or compliance statements from chip makers. So, AI can be really powerful in scaling how you can effectively pull that information out of the tens or hundreds of millions of parts that are out there on the market, do that efficiently. The other big area we look at is sort of when we're deduplicating and cleaning these big data sets, right? So, a given part might be listed by a bunch of different manufacturers, or by a bunch of different distributors under similar, but not identical names, maybe slightly different part classifications, slightly different manufacturer names, things like that. Just, you know, there's been a lot of mergers, and acquisitions in the chip making space over the past 20 years. And so, you see a lot of this kind of like messiness where no one's really wrong, but if you put all of their rightness next to each other, it doesn't quite line up and make sense as one central truth. And so, AI can be really powerful in helping to spot those trends and take these data sets, and figure out like, oh, when this chip maker or in this distributor says this chip is made by this company, and this distributor says it's made by this company. They're talking about the same thing. And now we can spot that pattern and we can take that data and merge it together, and understand when their data is, you know, what the whole truth is for a given part.

Phillip: And beyond that, there's a practical effect sort of to the customer, which is you'll have a whole body of users of this data ultimately from the engineer onward. And it goes down not only to the sort of complex, what is the market, but really simple information like, what is a description of this part? What type of thing am I supposed to be looking at? What is this thing? What is this object? And even those-

Matthew: Are we even talking about the same part? Yeah. We've seen folks go back and forth for a while saying like, I approved this part. You got this part. And then at some point they realize they're talking about the same part. But, you know, the MPN just looks a little bit different. And so, like, it's really useful for just spotting those kinds of things and getting everyone on the same ground truth to say like, okay, if you say one, two, three, and you say one, two, three, four, you're really talking about the same thing. And you can all, you know, breathe a sigh of relief and move on with production.

James: Do you think it's going to get to a point where it's going to be impossible to actually compete in the space without using advanced AI?

Matthew: Yeah, I mean, I think, as with any of these things, it's going to take a while. So, I think it's going to be progressively harder to do a good job or to do as good a job as you can do if you're not leveraging the latest tools. And that's usually the case. And so, I think we'll start to see folks who are maybe doing some of these data cleanup approaches using humans. That's going to become really hard for them to scale. And so, the folks who are using the latest, and greatest technologies to enable the efficiency of these processes are going to start to have better quality data and pull ahead. And the OEMs and the contract manufacturers that leverage those tools are going to find themselves kind of increasingly able to out-compete their competitors, and deliver products more effectively and at a lower cost. So, I think it's not going to be like tomorrow, anyone who's not using generative AI, or whatever the latest buzzword version of technology is are going to be out of business. But I think as with any of these things, if you kind of remain stuck in old ways of doing things, eventually that's gonna become a competitive disadvantage for sure.

Phillip: Yeah, it enables a lot of opportunity to do the intelligent labor when the intelligent labor is needed, right? Joseph, you brought up sort of the question earlier. What does it mean for a procurement team specifically? And when we think about ongoing production of a product and you're thinking about hundreds, or thousands of components over multiple releases over multiple months, or years to get the best scenario to get the best financial outcomes for your organization to know when you should maybe move between distributors, or move between manufacturers because there's opportunities on the market that can lower your unit economics, and then pass that value onto your customers. That type of really diverse data, right? Like there's automation that can just sort of give you that opportunity, and present it to you in a very proactive way so that you can go after as a leader in the procurement space, a much more strategic opportunity and really like dig deep into the stuff that does need that sort of human touch. And I think that's really, that's the outcome is that, yeah, you could probably still, you know, build a circuit board without a bunch of AI and automation, but are you going to have the best manufacturing and the best financial outcomes on that circuit board without them? Certainly not.

Matthew:  Right. I mean, look, it's the same as designing a circuit board. Technically, you could still tape out your PCPA, and make your bare boards off of that. And people did it for decades. But is that the most efficient way to do it? Is it going to allow you to build the most sophisticated products? Obviously, anyone who's listening to this knows the answer is you're going to want to use eCAD software. I think it's similar to that. Obviously, much earlier days, though, in that transition.

Joseph: Do you think this is the end of the data entry specialist?

Matthew: You know, I think data entry as a job is probably gonna be on the decline for a while. We use a lot of AI to help with these kinds of like data extraction problems. And it's certainly not perfect yet. You know, for example, we use, you know, machine learning and AI to extract relevant data out of, you know, shipment notifications and invoices and packet slips and all that stuff from distributors. And like, I will say using the latest and greatest AI, it still regularly makes mistakes that require human intervention. Now it does 99% of the work, and the humans can do 1% of the work instead of the humans doing a hundred percent of the work. So that's not to say it's not of a great deal of value, but it's not a hundred percent solved. I suspect though, that if we look even just a couple of years into the future, that that wedge is going to start to significantly narrow. You know, a lot of the materials we deal with, data sheets, things like that, are like really rough documents, right? A lot of these were designed a long time ago. You know, I joke that some data sheets are barely human readable, let alone machine readable. But the rate at which some of these kinds of like extractive technologies are advancing is pretty incredible. What was completely unthinkable and when we tried to do it a year ago, totally impossible, is now sort of like every day. And so, I think if you look a year or two forward, I suspect those kinds of tasks of just pulling stuff out of documents is going to be a pretty software-driven, and AI-driven workflow for sure.

James: So, before the podcast, we were actually chatting about the history of the company and you were mentioning that you have had quite an evolutionary journey over the years, changing functions. Do you want to tell us a little bit about where you started and how you got to where you are now?

Matthew: Yeah, you know, Phil and I, our journey sort of starts together before Cofactr. Both of us, and actually some of our team, originally founded and ran an engineering services firm that did kind of weird experimental prototyping work at the kind of like engineering design intersection. We worked on theme park attractions, essentially, and self-driving cars and all sorts of things from sort of an engineering design perspective. And after we sold that company to private equity, we felt like we had a lot of kind of like challenges and maybe frustrations with how our experience of the electronics manufacturing space had been. And so, we went off and started a quick turn contract manufacturer, and started building a bunch of software to help power that. And that was super valuable in terms of giving us all of the experience out of being on the other side of that and discovering all of the fun challenges that contract manufacturers have. And so, it was sort of out of those two sets of experiences that Cofactr ultimately evolved as it exists now. But yeah, Cofactr started off as a quick-turn contract manufacturer, running, you know, picking place lines and all that. Pretty quickly, though, we realized actually doing the manufacturing side was not maybe that exciting to us. And also, in fact, existing manufacturing manufacturers are quite good at that. And we weren't really bringing a lot of innovation to the how do you put chips on a board space. But all the software we were starting to write and inventory management systems we were starting to develop to make sure the chips were ready to be put on the board was exciting people. And so, we, you know, we're nothing if not good listeners. And so, we, you know, followed, followed, followed that to where we are now.

Phillip: Yeah, it gave us a lot of guidance and what we should focus on as a company. Like just having the hands-on experience of having designed and then built circuit boards, like what we should maintain, what we should maintain ownership of as an organization, what we should continue to do. We never had this sort of pivot to just a pure SaaS data model. It's hugely important to what we do. It is the underpinning sort of fabric that everything is built upon, but we have a physical warehouse. We are buying parts. We are receiving those parts. We are counting those parts because we know that there's no amount of data that can clean up the sort of reality of dealing with physical materials. And ultimately, that's such a burden for an OEM, or a CM doing quick turn work, or sort of numerous other customer profiles that we could potentially interact with. And I think that it's all that sort of like messy, weird stuff that turned into more organized, standardized, but ultimately manufacturing processes has left us with a business model that's pretty unique in where I think our hands still, the perception is we are willing to get our hands a lot dirtier than I think a lot of other organizations sort of in a similar space.

Joseph: So, which pain points is Cofactr? Sorry. Which pain points is Cofactr particularly well-placed to help clients deal with?

Matthew: Yeah, I mean, so I think there's some pretty obvious ones for, like, engineering teams and new product introduction folks, right? So, we work with a lot of big, big OEMs that people have probably heard of that, you know, on their R&D side. And so when you're in that kind of rapid prototyping phase, you doing a lot of boards you're buying a lot of parts, and what we see is like engineers just end up spending like upwards of 20 of their time dealing with getting their prototypes made instead of sitting in a product like Ultium actually designing those boards, and you know doing the job of an electrical engineer, and so that's an easy one. We work with those folks to essentially outsource all their kind of like pre-scaled production, inventory management and procurement into a software platform, let their engineers, you know, sync their BOMs over from 365 into Cofactr, click a button and kind of just go back to their day. And so that's like a really easy one. And, you know, folks on those kinds of teams tend to love that. As you move into the scale production it becomes really about how do we understand risk how do we think about the future of this product right. Now that we're working with what our engineers gave us essentially are we making the right like kind of buying decisions, and optimizing our procurement strategy across all distributors, so it's these kinds of like just in time-time versus just-in-case pain points are really important. And I think it also flows back the other way, which is that like teams that use co-factor, it enables the engineers to make more supply chain aware decisions early on, right? I think like it's not necessarily on engineers to be supply chain experts, but more and more we see that like if they're not making the right part selection choices, and the right alt selection choices, day one just makes more work for everyone later, right? And engineers don't want to have to re-spin boards any more than their employers want them to have to re-spin boards. And so, I think it's also about flowing those kinds of supply chain considerations back into the engineering process. So, we really try to sit at that kind of like intersection between engineering teams, and supply chain folks to solve these kinds of like messy data and operational problems for them in a way that ideally feels like pretty effortless.

Phillip: And maybe I'll keep harping on the physical stuff here. There's also just, it's a pain to manage these materials for CMs, for EMS providers. You might have a great relationship with a handful of distributors that are sending you materials exactly as you need them. But co-factor can work as really an aggregator of diverse materials, standardizing those materials and the actual process that leads to them arriving at a manufacturing line ready to go. And that's everything from, are they being reeled to the specification of the actual PCBA, right? Like, are they labeled correctly? Do they have the right barcode so that I can receive those easily? I mean, all these things.

Matthew: The traceability, right? Yeah.

Philip: Right. There's just a bunch of stuff that you sort of say well how do you successfully receive a material, and then get it to the line without it going through a couple of departments, and that's the kind of stuff where you look at Cofactr, and you go oh it's all just it's all just one thing, and it all just kind of makes sense now, and so there's the there's the OEM pain points, but cm's is a totally different animal.

Matthew: And ultimately that's kind of just like, from our perspective, that's like table stakes to do this stuff well, right? If you plan the most amazing, beautiful supply chain in a spreadsheet, but then it gets screwed up when you go to execute it in the real world because the data doesn't match the reality, or because you ordered it, but it goes missing somewhere along the way, or you got it, but you lost all the traceability paperwork and now you can't use it. Like that spreadsheet is sort of no use, right? And so, it's really crucial that what happens in the software then also happens in reality, and kind of vice versa. That's the other thing is like we've seen folks really get tripped up by building plans based on faulty inventory data. It turns out what they had in their ERP didn't match what they had in their warehouse. And those were, you know, long lead time parts or expensive parts. And it can really jam up production in sort of big consequential expensive kinds of ways. And so, for us, it's like, we want to make sure people are doing all this kind of like highfalutin risk management, cool data stuff on a foundation of just like actually accurate information. And, you know, one of the inputs to that is just like, what do you own? Can you use it? Is it where you think it is? And so those are those are like, unfortunately, not kind of like cool problems, but like absolutely critical to be able to do anything else successfully.

James: Being spread across so many trends, being so aware of everything that's happening in the space, what do you see happening in the area of reshoring and nearshoring? Is it as extensive as people are saying it is? Or is it sort of a little more toned down and being hyped up?

Matthew: Yeah, I mean, I think we probably have a skewed perspective on this in that our customer base is heavily in industries that are doing that, and maybe less heavily in industries like consumer electronics that maybe are doing it less. I don't know, because we just don't have that perspective on it. Certainly, among our customer base and what we'd sort of call critical industries like aerospace, defense, robotics, marine, automotive, industrial controls, I think we're seeing a strong trend. And, you know, those, those were the industries that were most initially inclined to near shore, or maybe for regulatory reasons, we're always on shore. But yeah, I would say that we definitely see folks who I think would have looked seriously at manufacturing overseas a few years ago, looking seriously manufacturing in Canada or Mexico instead, or trying to figure out, well, how do I do a supply chain in the US? It's certainly like, you know, in some ways easier said than done. There's a lot of really nice, convenient things about manufacturing in Shenzhen. But you also give up a lot in that process in terms of control and stability and predictability. So, I think like's a valid trade-off to mid. We're certainly also seeing folks making big investments in manufacturing in places like India, which is certainly not nearshoring, but is about diversifying where in Asia you're manufacturing if you're manufacturing in Asia. So I think it's some of both, but certainly we see a lot of folks working on building plants in Mexico right now. And I think that's being driven by legitimate demand. I've never known EMS and CM type companies to spend money building facilities for fun. They tend to be pretty like market-driven participants. And so, like if they're building factories there, there's probably-

James: I've seen some trends in a lot of factories opening in Vietnam, Indonesia, and Malaysia as well.

Matthew: Yeah, that's definitely something we see. We have a handful of customers who manufacture in Malaysia and seem to be doing so successfully. So, I think that resonates with what we've seen for sure.

Phillip: Yeah. The opportunity there seems mind-blowing right now Like there's definitely a number of industries that have already sort of seen the, you know, blank blank for this part of the world already happen. And the interest that we see specifically from India is massive on supply chain tools, and procurement automation. It seems like it's a real moment of opportunity in that market. And it's fascinating to sort of be sitting so front, and center for what is like a giant global shift in the way that we're thinking about traceability, and reliability and just caring so deeply about where your stuff gets made. It's going to be quite a decade.

James: I think part of that.

Matthew: Yeah, I mean, I think I think in general, say i think in general like the interesting thing with manufacturing right is like contract manufacturer ems really do two kind of separate things one is like run factories that take all these materials, and put them on circuit boards, and it's comparatively easy to like put those in different places right you know you can ship a pick, and place to any country. And, you know, generally speaking, you can hire staff and train them. And, you know, some markets are cheaper than others, but that's sort of like pretty manageable. But then the other thing they do is like manage supply chains in a lot of cases. And supply chains look just fundamentally different in China than they look really anywhere else in the world, particularly for things like electronics. And so that's where I see the big adaptation happening. And Phil was talking about this with some of the conversations we've had with Indian contract manufacturers, where they're saying, look, I built the factory in India. It wasn't that big of a deal necessarily for me to put up a building and get people to work in it, and put these machines in. But like, I'm not in Shenzhen. Like the massive sort of like electronic parts markets and infrastructure of brokers, and independent distributors, and manufacturing sites and, you know, people who just know how to make those supply chains run efficiently don't exist there. They arguably don't really exist in the same way in the US or really anywhere other than China, right? And so, I think you see that the construction of these supply chains has to look somewhat different in all these other countries. And that's, I think, where the biggest kind of like learning curve, and adaptation curve is having to happen is figuring out just like if I can't just tell my CM magically source it all from down the street. How does that work? And that's always a challenge for sure.

James: What do you sort of see the role of these new legislations about traceability and supply chain coming into play? What role are they going to play in the shift as well? Because I imagine that it's going to be a lot easier for some of these companies to shift to these areas where it's easy to justify that they're above board, especially with European regulations and what they're putting in place with supply chains.

Matthew: Yeah I mean, I think that's right, you know. Generally, there's one of the good and the bad thing about manufacturing in China is it tends to be kind of a black box, right? And that's good if the black box is spitting out the right output that you want and you don't really have to worry about it. It's bad if people are expecting you to know what happens inside the black box or if the black box, you know, isn't doing what you want because of supply chain disruptions, right? And so, I think that's correct, that, like, fundamentally, we're going to see a lot more of that. You know, the way that parts find their way to the factory floor can be a really straight line. You know, we talked to some customers where it's like, I call up my semiconductor manufacturer and I order $20 million worth of some chip, and then they ship that chip some number of months later to the CM, and it's all pretty clean and tidy. It can be quite meandering, involving all sorts of authorized and unauthorized distributors. And there's no value judgment on whether one of those is better or worse. But certainly if your if your requirement is like traceability, whether that's because you're in a regulated industry, where that has always been the requirement, or whether, you know, that's sort of a new requirement for you, because of evolving regulatory regimes, people are going to have to figure out how to reengineer their supply chains to fit within those rules. And certainly it's going to be easier to do that in, in manufacturing areas that you're building supply chains a little bit more from scratch or, or that are, you know, like the U S and parts of Europe, more simplistic supply chains, typically where it kind of goes like, you know, manufacturer, distributor and customer versus places like China, which sometimes can be a bit more, you know, convoluted in terms of how the parts, you know, make their way to where they're going.

Phillip: There's quite alot of pink plastic bags with part numbers written on them and pen in the world. And I have a feeling that like we're definitely doing as much as we can to turn that into a beautiful printed label with some traceability, and a COO. But I think that there's probably over the next few years, you're going to see so much automation, so much software, so much traceability that just like really make sure that that audit trail is auditable and trailable. And it's fascinating, especially in contexts where a manufacturer might handle the NPI process, but then it moves into scaled production. And at scale production, you might have full traceability on your components, but a few of those components from the NPI process, if they bleed into the production environment, you might all of a sudden sort of have questionable materials landing on boards that have sort of universal traceability. Generally speaking, it's a really fascinating world to sort of stumble into over the last few years and try to make a positive impact in.

Matthew: Yeah, and that's also a pain point we very much solve. I think we hear from contract manufacturers a lot, like more and more they're being told by their end customer, we need this kind of end-to-end traceability. But as a practical matter, it's really hard to actually do it. It's just a huge administrative burden to keep track of all this paperwork and make sure that you haven't missed one out of the hundreds, or thousands of parts in the design that you need that for. And so, again, that's an area where like software and automation is crucially important to meaning that like a contract manufacturer can, can meet those requirements without having to hire like 200 people just to like, you know, catalog certificates of conformity and, you know, keep all that straight.

Joseph: Do you have any insights you could share about adapting to world events that disrupt the supply chain?

Matthew: You know, I mean, I think like the key things, of course, are designed for flexibility to begin with, right? And so, this is a big shift we've seen over the past few years driven, you know, by the COVID-related disruptions is looking back a few years, people would like pick the part they pick, and they'd send it to the supply chain team. And that was kind of the end of it, right? More and more we see engineering teams who were burned by that experience thinking about these considerations more upfront, whether that's picking parts that have a bunch of different alternatives across a number of manufacturers, even if maybe a different part would be like technically a little bit nicer for them, but, you know, would increase the risk of the sort of product having supply chain disruptions. We've seen folks designing multiple footprints in, in some cases. So, if the board isn't super space constrained, maybe you design to support a given part, but, you know, handle a few different potential footprints of it. And that gives you a little more optionality. So, I think like we see the OEMs that are kind of most successful at avoiding these kinds of disruptions, leveraging techniques like that. I think another trend I've seen is just like people being more flexible, right? So like we have some customers that are building we have a customer that comes to mind that builds uh building rocket launch systems. And they have like a really deeply rooted kind of software-y approach to how they think about hardware engineering, where they try to really build very rapidly, rapid iteration. So, they're not worrying too much about, is the design I'm making today good for five years from now? They're just like, it just needs to work for this round. And we actually accept that we're gonna build our entire product with the idea that like respins are inevitable, and that's just part of our culture in the same way that like people build software products with this very kind of rapid iterative cycle. That's not an easy thing to do for an org if you didn't sort of like build from the ground up with that kind of ethos, but it's really successful in letting them move incredibly quickly and have a lot of adaptability. And obviously, the more you have modern ECAD software, and that's integrated with your supply chain systems like Cofactr, and you have kind of a diverse and flexible contract manufacturing base, there's a lot of ingredients required to do that kind of thing successfully. But it can be really successful as well. And then I think the other thing is just like being still willing to do a certain amount of just-in-case stockpiling, right? I think we're definitely not at a place yet where people can just switch fully back to just-in-time procurement and expect that to be a successful strategy. You know, I think, like I said, on the flip side, I don't think you need to stockpile every possible part you might ever need for any board forever, or you're gonna end up with a lot of chips on the shelf and a lot of capital tied up. But you know, I think it's about being really strategic about identifying which parts are potentially most likely to be subject to disruption and then being, you know, a little more conservative about putting those on a shelf so that you're not regretting that later.

Phillip: And this is a really practical sort of simple one, but a really healthy parts library, I think, is key to this. We've definitely heard examples where you've got a parts library, you've got 10 alts, you feel good, that feels very sourceable, but that the procurement process is often by the first one, first one's no good, by the second one. Now, is the health of the third through the 10th alt in that library under that internal part number, are those in good condition? Are you actually in a defensible position if you lose a couple of those alts or that primary part? And that's kind of what we're really practically helping with some of our customers is just what's the actual health of your library, right? Like, are you in a situation where if something falls short, you have plenty of backup and you know that you're going to still be able to manufacture and move forward. So that's one that it's really simple, but, you know, it's really easy to forget about the last two thirds of volts in an internal part number and just sort of shrug it off. And God forbid, they fall before the first ones do.

Matthew: Yeah, I mean, the TLDR on that is like, just keep an eye on things, whether you're doing it manually or using software to do it, right? Like you kind of just have to monitor. And I've seen this happen and really trip up some companies where the engineers are like, we're going to be really proactive about supply chain. We're going to look at Octopart or look at a distributor website. When we design our boards, we're only going to design with parts that we can buy. They design all those parts in that they can buy. They send that BOM over supply chain. It sits on a shelf for a few months because that board is in line behind a bunch of other boards. They go off to buy all the parts they need. And we could have gotten plenty of them a few months ago, but now we're out on two of the parts and now we're going to have a totally avoidable re-spin, right? We could have gotten that distributor to allocate that for us, put it on a shelf, not even bill us probably until those parts shipped later. So, it wouldn't even have been a cash flow problem. It's just about being proactive. So, just monitor what you actually need and what the sort of conditions those products are, so that the moment it starts to trend in the wrong direction, you're able to jump on it. And again, that's something that like humans can definitely do manually, but it's a lot easier if you have software helping you, especially as your part library goes from, you know, maybe a few hundred parts when you're an early startup to, you know, tens of thousands, hundreds of thousands of parts, you know, as a more scaled organization.

James: Last question here from us is, do you see any emerging technologies or approaches that are probably going to shape the next generation of inventory management practices?

Matthew: Yeah, I mean, I think there's a few key areas certainly like we were talking about earlier these kinds of like extract data extraction stuff is getting a lot better right, and so inventory management has traditionally involved a lot of paperwork and so a process we've seen a lot of folks have is you know boxes of come in, they have a whole pile of certifications and packing slips with critical compliance information on them. And those all just go in a file cabinet dated the day they arrived. And if they ever need to look at that information later, someone has to go, you know, rummage through those file cabinets for hours or days to pull all the right documents to like put it all together, right. And so I think, you know, scanning them as a start but if you but you know a bunch of like pictures of packing slips isn't really that much more useful than a file cabinet full of packing slips, but with you know better optical character recognition, and ai driven extraction techniques you're certainly going to see a lot more just like every shred of information that has to do with this piece of material is captured digitally is linked to the correct record is is you know available and searchable and reportable. And so, you know long before it's an issue of a recall, or an issue of a compliance audit, you know, hey, we have all the right paperwork. We know exactly what part went everywhere. And I think you'll also see a lot more kind of like really granular traceability like you see in the medical device space extend to other industries, right? So, like, you know, if you're operating under an FDA, you know, CFR 21 regime, you've always had to have this really granular, like exactly which chip that I buy, and which board did it go on? And where did that board go level of record keeping. And that's, you know, traditionally been kind of like, pretty specialized and a little bit burdensome for folks in that industry. But I think, you know, with these evolving European regulations, with evolving kind of norms, maybe in the US and some regulatory changes, you're going to see people need that in a lot more places. And kind of the only way to do that is with more digitization. It's just like, so, you know, insurmountably expensive and impractical to try to do that kind of stuff manually. And so, I think you'll start just seeing a lot more connectivity between the inventory management systems that contract manufacturers use and the tools that OEMs use. And obviously, we have a vested interest in saying this because we build those kinds of tools. But I think we build those tools because that's where we see the industry going, and where we see the OEMs, and CMs that are at the leading edge of trying to like make these processes better, looking for solutions like that.

James: Great. Well, thank you so much for coming on the podcast. It's been absolutely fascinating. I know I've learned a lot. I'm sure you have too, Joseph. But for anyone who's been listening to this and really been enjoying this topic, please stay tuned because we will be announcing in the near future a special webinar episode featuring Phil and Matthew, where they're going to go more in depth with these topics, discuss how they work with Octopart and Ultium, and basically just expand on what we've covered here today. So, Matthew and Phil, thank you so much. Really appreciate your time.

Matthew: Thank you for having us.

Phillip: Thank you, James, and Joseph.