Jason Busch Former CEO of Spend Matters - Will AI Make Procurement Obsolete? 的完整文字稿

master's. 00:03:54.560 --> 00:03:56.240 We called it a submatriculation. 00:03:56.400 --> 00:04:03.919 So I did uh I was born undergrad and then uh master's in engineering, and so I I skipped the uh the the liberal arts side. 00:04:04.000 --> 00:04:09.360 So between the two of us, we've we've got the full school covered except for nursing, but no locust walk. 00:04:09.439 --> 00:04:11.120 I mean, I I go back I all the time. 00:04:11.280 --> 00:04:17.439 I was in Philly a week and a half ago for a wedding, actually, and I made sure to stop by and get some good cheesesteaks. 00:04:17.600 --> 00:04:19.920 And you know, I've I have really fond memories. 00:04:20.079 --> 00:04:23.680 I really quite love the university, and I plan a lot of our reunions. 00:04:23.759 --> 00:04:28.240 I'm still involved, and uh and I really was close with Amy Gutman back when she was still the president. 00:04:28.399 --> 00:04:31.600 I thought she was just a such a phenomenal leader of the university. 00:04:31.680 --> 00:04:35.600 But you know, I'm I'm yeah, I'm I'm just I I looked at your LinkedIn and I noticed it. 00:04:35.680 --> 00:04:37.600 I was like, how have we never talked about this? 00:04:37.680 --> 00:04:39.759 It just was like such a missed overlap. 00:04:40.079 --> 00:04:42.399 Were you in a fraternity at at the university? 00:04:42.800 --> 00:04:44.240 No, that was not for me. 00:04:44.480 --> 00:04:46.560 I was uh a lot less fun back then. 00:04:46.639 --> 00:04:50.480 I I was focused a lot on uh studying, on research. 00:04:50.639 --> 00:04:53.759 I had the chance to edit a bunch of books when I was in college. 00:04:54.000 --> 00:04:55.680 I got in trouble a lot too. 00:04:55.839 --> 00:05:01.199 I I was I was an editor of of the then libertarian conservative paper, The Red and Blue. 00:05:01.360 --> 00:05:10.879 We we had our offices rated, you know, it wasn't the school somehow didn't like people speaking, speaking the truth if it was not the progressive, dogmatic, academic truth. 00:05:11.040 --> 00:05:12.800 And so I I did a lot there. 00:05:13.120 --> 00:05:21.519 My graduate advisor, who also oversaw my thesis, was a guy named Bob Engs, and he introduced me a lot to working with primary and material. 00:05:21.680 --> 00:05:23.920 Bob died 10 or 15 years ago. 00:05:24.160 --> 00:05:25.040 Fascinating guy. 00:05:25.120 --> 00:05:29.279 He was actually the one responsible for telling me to not go on in history. 00:05:29.439 --> 00:05:30.240 I loved it. 00:05:30.399 --> 00:05:43.199 I think I had a knack for it, but it was uh it was a time when if you are a white male, and God forbid a a white, you know, libertarian conservative male in the academy, god forbid you were also Jewish, it was not a good job market. 00:05:43.279 --> 00:05:45.759 I suppose you could go to Hillsdale or a place like that. 00:05:45.920 --> 00:05:50.319 But uh he joked around with me and said, you know, someday you can endow a chair for me. 00:05:50.480 --> 00:05:53.360 We'll get your ass out of here and uh get into business. 00:05:53.600 --> 00:05:58.720 So I that started my path into consulting and then the tech scene in procurement. 00:05:59.040 --> 00:06:13.920 But it was uh, yeah, it I think today, I think the problem for so many students is that they have this view of college that, you know, you party, uh you might have been a member of a fraternity, you know, you work hard, you play hard, you do a sport, you know, whatever it is. 00:06:14.240 --> 00:06:25.199 But I think you know, the ultimate experience, even today, especially today, when we're not sure what the job market is going to be for grads in the future, is just to make it your own. 00:06:25.439 --> 00:06:33.360 So through Bob, I got to work with a ton of Civil War source material, you know, reams and reams of letters from the front. 00:06:33.519 --> 00:06:35.199 And it was just fantastic. 00:06:35.360 --> 00:06:37.040 I proofread a few books. 00:06:37.199 --> 00:06:39.920 I worked, my undergraduate advisor was Drew Faust. 00:06:40.079 --> 00:06:44.000 I I think I might have, I think I might have proof one of her books as a research assistant. 00:06:44.160 --> 00:06:46.800 So I I was much more an academic back then. 00:06:46.879 --> 00:06:49.439 I was not as much fun as I am today, hopefully. 00:06:49.759 --> 00:06:52.879 But um, yeah, it was uh just just a great experience. 00:06:53.040 --> 00:06:54.800 I was very lucky, lucky to go there. 00:06:55.199 --> 00:06:55.839 That's amazing. 00:06:56.000 --> 00:06:57.519 No, it is a great university. 00:06:57.600 --> 00:06:58.160 And it's funny. 00:06:58.240 --> 00:07:04.480 So Drew Faust, just uh add a little bit of additional context, was the the president when I started my MBA at Harvard. 00:07:04.560 --> 00:07:07.680 And so yeah, there's a lot of there's a lot of sort of funny small connections. 00:07:07.759 --> 00:07:12.399 And so and so Robert Eng's focused on uh civil war and reconstruction. 00:07:12.480 --> 00:07:13.920 Was that also your focus area? 00:07:14.240 --> 00:07:14.639 It was. 00:07:14.800 --> 00:07:15.199 It was. 00:07:15.279 --> 00:07:24.800 I mean, I've always loved world history, tenants and pretty stringent requirements for languages if you were going to stop side the US and my language skills are quite horrible. 00:07:24.959 --> 00:07:28.560 I by default had to do American history, but I I thoroughly enjoyed it. 00:07:28.720 --> 00:07:34.399 So, and I find um, you know, now living through the period we're living in, which is absolutely crazy, right? 00:07:34.480 --> 00:07:39.360 We have two political parties which don't at all resemble what they were even a decade ago. 00:07:39.519 --> 00:07:51.839 We've had the hot swap of the parties, and you know, these things have played out in history to some degree as well, but you know, we're living through a period of of history right now, and I don't think a lot of people in it kind of see that. 00:07:52.000 --> 00:08:09.360 I I do think

having an appreciation for what's happened in the past and not jumping to conclusions until you have a lot of source material and data can really help you put things in perspective as well as a builder, but also to understand the implications of what you're doing. 00:08:09.519 --> 00:08:13.279 And you know, I've been five followed my passions at different points in life. 00:08:13.439 --> 00:08:27.040 You know, when I was at Free Markets, which was the first tech company I worked for after consulting, and I just did a podcast with the founder, one of the co-founders of it, Sam Kimmy, interviewing him on what we build and why and all of that. 00:08:27.279 --> 00:08:30.160 I, you know, I was not level-headed when I joined Free Markets. 00:08:30.240 --> 00:08:35.039 I was I I was passionate and you know, I was a zealot for what we were doing. 00:08:35.279 --> 00:08:42.720 We we could see at the time the notion of reverse auctions and creating competition where it didn't exist in the form to me was really valuable. 00:08:42.879 --> 00:08:46.240 But the lasting legacy of it had nothing to do with auctions, actually. 00:08:46.480 --> 00:08:59.679 Auctions were somewhat of a bad in the end, but it was about creating level playing fields between different suppliers, having better data flow back and forth, something I know you're doing at Light Source with those material material and other data. 00:08:59.919 --> 00:09:02.000 Back in the day, it didn't have level playing fields. 00:09:02.159 --> 00:09:04.480 Incumbents had far more access to information. 00:09:04.879 --> 00:09:18.320 In fact, a lot of a lot in lots of companies and manufacturing, I mean, even today, a lot of bids are rigged, not financially so, but design engineers want to, you know, see certain suppliers get the awards. 00:09:18.480 --> 00:09:24.879 And free market's ultimate legacy to me was we were really good at specifying requirements and communicating them. 00:09:25.200 --> 00:09:26.320 Yeah, tell me more about that. 00:09:26.399 --> 00:09:28.480 What do you mean by specifying requirements? 00:09:28.799 --> 00:09:46.799 In the past, you know, if you had uh, and and I'll I'll put it in the old-fashioned speak, you know, if we were sourcing 200 printed circuit boards, we might create, you know, seven or eight seven to ten lot structures or groupings of printed circuit boards based on you know logical bunches of suppliers who could bid on them. 00:09:46.960 --> 00:09:50.480 And or you might create a super lot as an alternative. 00:09:50.720 --> 00:10:02.000 But in in the past, you know, the incumbent suppliers may have had, it may or not have been privileged information, but they had a greater understanding of what it talked took to build those PCBs or PCBAs. 00:10:02.159 --> 00:10:07.919 And you can swap out a metal casting or forging or a finished component in automotive, it doesn't, you know, matter. 00:10:08.159 --> 00:10:15.440 And often what happened was, you know, you had design engineers make decisions which were not based on trade-offs, and that limited your supply base. 00:10:15.600 --> 00:10:26.639 And if you wanted to expand it through making slight tweaks or variations or expand it geographically, there would often be pushback in in the engineering errorization, and procurement had to come in after the fact. 00:10:26.960 --> 00:10:31.039 So what we did, we were really good at free market communicating specifications. 00:10:31.120 --> 00:10:37.039 So we we would create literally these massive bid packs, these were like this big, and we would nail them out. 00:10:37.200 --> 00:10:40.480 It was before the day of giant files, you know, you could send. 00:10:40.639 --> 00:10:42.879 We would send out uh design drawings. 00:10:43.039 --> 00:10:49.759 Uh, we would have uh we called the market making engineers answer engineering questions from suppliers. 00:10:50.000 --> 00:10:54.559 So by the time bid day came around, suppliers were incredibly well prepared to bid. 00:10:54.720 --> 00:10:58.720 They knew exactly what they were bidding on and they knew how far they could go. 00:10:58.879 --> 00:11:01.440 And we did that to enable the auction. 00:11:01.519 --> 00:11:06.799 But I think the irony of all of it is that the lead up to the auction is what enabled the outcome. 00:11:06.960 --> 00:11:30.879 And even though auctions got a bad rap, I think because a lot of people use them incorrectly, the process we went through to enable the auction became really the basis of not only modern, you know, direct material sourcing of technology, but even the extension of PLM into you know direct materials collaboration solutions. 00:11:31.120 --> 00:11:36.399 That to me was the legacy of that company, besides spreading free markets, which is what what what we were about. 00:11:36.720 --> 00:11:40.799 Yeah, why do you feel that e-auctions did not persist longer term? 00:11:41.039 --> 00:11:53.440 You know, you you called it a fad, and you know, I I tend to agree with that perspective, but I I feel it's some ways there are certain businesses that uh, you know, because we get asked about e-auctions all the time and you know, whether or not we do them. 00:11:53.600 --> 00:11:58.000 My response is usually to ask the question, like, well, how how many e-auctions are you running? 00:11:58.159 --> 00:12:05.279 And then nine times out of ten, the response will be, well, none, but you know, we want to have the capability to do it if we want to. 00:12:05.440 --> 00:12:16.320 Um, but then there are one out of ten businesses like, you know, BASF or Walmart or Amazon, where they're actually running like a good number of auctions, maybe just because of their scale or the kinds of commodities that they're buying. 00:12:16.480 --> 00:12:21.200 So I'm kind of curious, you know, why has the why have e auctions gone the way of the sort of dodo? 00:12:21.600 --> 00:12:31.279 I think when auctions stick, and you referenced some organizations with a long

history, I think, of successful direct materials procurement, Amazon being more recent, but being a very strong organization too. 00:12:31.440 --> 00:12:36.720 So if you look at the auction itself, you know, you can think of it, we we gotta go back to calculus and waves. 00:12:36.879 --> 00:12:41.519 It's like I'm gonna use the incorrect mathematical term here, but it's the crescendo of the wave, right? 00:12:41.679 --> 00:12:45.679 So all the work occurs before and after to make an auction work. 00:12:45.919 --> 00:12:53.440 The issue with auctions in so many companies is they had no idea that all the work, 99% of the work was the preparation and then the implementation. 00:12:53.600 --> 00:13:04.639 So if you don't prepare properly and you don't implement properly, you may identify the savings, but they may never become realized savings or they and or they will get eroded over time. 00:13:04.879 --> 00:13:06.799 A lot of people looked at free markets. 00:13:06.879 --> 00:13:19.840 So I could go back to old names in the field, you know, Commerce Bid, you know, within Commerce One, Procuri and Taurus, all these companies started building self-service technology that, you know, copied a lot of what we were doing. 00:13:20.000 --> 00:13:29.120 I remember at free markets we lawyered up because we had a ton of IP around different auction formats, and we would sue some of these people for infringement at the time. 00:13:29.440 --> 00:13:37.279 But what they did is they put that IP into self-service tools, which didn't require what we would call full service or market making at free markets. 00:13:37.519 --> 00:13:45.440 And so what happened was organizations skipped the preparation steps and they skipped the post-auction steps to drive successful implementation. 00:13:45.679 --> 00:13:55.279 So therefore, auctions got a bad name in lots of companies because on bid day you might get five, 10, 30, 40% savings, but it would never become realized. 00:13:55.759 --> 00:14:05.200 So, you know, there was one one other company like Free Markets, which was lost to the annals of procurement history abbreviate, which was a spin out of AT Carney. 00:14:05.440 --> 00:14:08.960 They did some of this successfully because Carney was services driven. 00:14:09.200 --> 00:14:23.279 A lot of the work that is now done today that's successfully implemented with negotiations is still done by large consultancies on direct materials simply because whether or not there's an auction, there's tremendous preparation. 00:14:23.440 --> 00:14:30.080 And most procurement teams for direct spends still don't go through that level of prep necessary to get the most out of an option. 00:14:30.320 --> 00:14:30.480 Right. 00:14:30.559 --> 00:14:31.120 It makes sense. 00:14:31.279 --> 00:14:35.759 I mean, look, this is a lot of the reason why you know we're focused on that PLM kind of connection. 00:14:35.840 --> 00:14:39.679 And if you've read the spec to scale manifesto, that's the first step, right? 00:14:39.759 --> 00:14:45.440 Is the spec alignment and just being able to even describe to the suppliers what it is that you're asking them to quote. 00:14:45.600 --> 00:14:48.000 If you can't do that, then they can't provide you with bids. 00:14:48.159 --> 00:15:04.559 And so, you know, the problem is if you don't have your data, you know, ducks in a row, that work of doing the the preparation for the event becomes like a monumental transformation task versus if you can actually have software and tooling that makes all of that information available to you, uh, then it can be be done relatively quickly. 00:15:04.720 --> 00:15:08.399 The interesting thing about an auction is I think the precision also just has to be a lot higher. 00:15:08.559 --> 00:15:12.639 Because if the preparation is done perfectly, then the auction can run smoothly. 00:15:12.799 --> 00:15:29.200 If the preparation is done, but then there's blemishes, or like, oh, that part doesn't have a drawing, or you know, that specification isn't clear, uh, you don't have a lot of opportunity or time to fix it before the auction starts to fall apart and people can't participate versus an RFQ where you're iterating multiple rounds. 00:15:29.360 --> 00:15:37.440 There's sort of like, I think about it sort of mathematically also like an RFQ, as an RFQ's number of rounds approaches infinity, you know, it becomes an e-auction. 00:15:37.679 --> 00:15:38.399 You know what I mean? 00:15:38.559 --> 00:15:42.639 It's just it becomes a real-time auction if if you have enough rounds and enough bid feedback. 00:15:42.720 --> 00:15:51.039 And so the line does get a little bit blurred, but I do think of RFQ as like an e-auction, but with a little bit bigger uh margin for error on the preparation side. 00:15:51.200 --> 00:15:56.080 You know, it doesn't fail because you can you can hustle people along without it being like time's up, you know? 00:15:56.320 --> 00:15:56.720 Totally. 00:15:56.879 --> 00:16:17.759 And I've become, you know, I I'll I'll say from a geek perspective, enamored over the years of essentially taking sourcing optimization technology, you know, with constraints and solvers and running scenarios, and then also being able to provide feedback to suppliers to adjust their bids competitively based on the constraints you might be applying. 00:16:18.080 --> 00:16:19.919 A few providers who've gone down that path. 00:16:20.080 --> 00:16:33.039 I know you know we're even over in overlapping spaces today at my startup gain, but you know, this the work that, and and I'll just, you know, give credit to where credit was due, CombinedNet started out doing a lot there. 00:16:33.200 --> 00:16:35.200 It's now you know owned by Jagger. 00:16:35.279 --> 00:16:43.840 There hasn't done a lot with that asset in terms of RD relative to what Koopa's done with ASO and Kilvar, um, and lots of others. 00:16:44.080 --> 00:16:48.240 So I I think that is a fascinating area for Direct as well.

00:16:48.559 --> 00:16:50.639 But again, I I think you're right. 00:16:50.799 --> 00:17:00.720 I think you go back to bridging that gap between design engineering through PLM specification, sharing, and then just you know, the rapid iteration and back and forth process. 00:17:00.879 --> 00:17:07.519 So you know, bear in mind we had lots of different people involved on the procurement side, and then we have different people involved on the supplier side. 00:17:07.680 --> 00:17:17.599 So, you know, whether it's a massive CapEx project, you know, to build out a battery factory, or whether it's you know, a wire harness, it's still different groups of people involved. 00:17:17.680 --> 00:17:20.720 So that collaboration piece becomes totally key. 00:17:20.960 --> 00:17:29.920 And I think too, you know, as well, where we can both get into with how AI evolves a bunch of this, it's a huge efficiency gain. 00:17:30.160 --> 00:17:36.720 And in my new startup, we're driving to autonomy for a lot less complexity compared to what you're doing at Light Source. 00:17:36.880 --> 00:17:42.799 But the the notion is can you build autonomy with or without human in the in the loop depending on the organization? 00:17:43.119 --> 00:17:44.640 Still, you need good data. 00:17:44.720 --> 00:17:50.000 You need to ensure it's accuracy, you need to keep overcommunicate that with suppliers all the time. 00:17:50.160 --> 00:18:01.039 You need to be able to answer questions, you need to remove doubt, and then you know, you need to remove any any latency in the process, whether that is pre-bid, post-bid, implementation, and the like. 00:18:01.200 --> 00:18:04.240 So I still think it's just the golden opportunity. 00:18:04.400 --> 00:18:07.119 I don't think the world needs another e-sourcing tool. 00:18:07.279 --> 00:18:20.880 There's been a few of those over the years, but again, if you think about sourcing as kind of that crescendo, it's everything that comes before and everything that comes after that makes the beauty of that negotiation a great outcome. 00:18:21.200 --> 00:18:22.240 Yeah, it's interesting. 00:18:22.400 --> 00:18:29.920 Well, how do you think about applying this mindset to something that is more of uh an RFP than an RFQ? 00:18:30.079 --> 00:18:37.920 Because I think if you have perfect information that you're sharing with suppliers, okay, you know, it makes much more sense that you can have competitive events. 00:18:38.160 --> 00:18:49.839 But a lot of times what we find is also that folks will say, okay, we're not at the scale where we're going to own every single part of our design, or maybe it's just a service and actually it doesn't have a SKU associated. 00:18:50.000 --> 00:18:55.039 I mean, how do you keep RFPs really competitive when it's really a proposal that's coming from the supplier side? 00:18:55.359 --> 00:18:56.319 It's a great question. 00:18:56.480 --> 00:19:00.480 I um I will answer that, but uh I'll take it in a slightly different direction.

00:19:00.720 --> 00:19:02.960 So I was obsessed at free markets. 00:19:03.039 --> 00:19:07.599 I nearly pulled off an acquisition when I was in corporate development 27, 28 years ago. 00:19:07.759 --> 00:19:10.079 I wanted to acquire a company called Hologix. 00:19:10.400 --> 00:19:13.599 It turned out we ended up using it in a commercial pilot. 00:19:13.759 --> 00:19:15.359 We couldn't get it to work as we wanted to. 00:19:15.519 --> 00:19:28.480 But the idea was, and we were focusing on much more basic stuff than you guys do today primarily, and certainly we do again, but you know, matching, for example, uh a fabrication shop's equipment with potential outputs. 00:19:28.559 --> 00:19:34.799 So they have a certain number of machines and a certain skill set, or you take a fasting conjecture and molded shop and their equipment. 00:19:35.039 --> 00:19:40.400 So what theoretically could they produce based on that equipment, not that they're producing today? 00:19:40.559 --> 00:20:00.480 Or maybe you bring in, in the case of metals, even a third party to provide a finisher coating after, because essentially, I mean, if you look at a metal part, you know, you're you essentially have input, you somehow trans transform the part with heat or with pressure or by cutting, and then you can code it, you you can put it into an assembly. 00:20:00.640 --> 00:20:01.920 There's only so much you can do with it. 00:20:02.079 --> 00:20:05.839 So theoretically, what can a machine do and what can a shop do? 00:20:06.160 --> 00:20:16.559 And then you you take a defined buyer requirement, a part specification, and can you map a part or a finished product specification to the machine capability of suppliers? 00:20:17.039 --> 00:20:18.640 I was utterly fascinated by that. 00:20:18.720 --> 00:20:22.079 I worked with with Sam Kinney and a few other people to see if we could make it work. 00:20:22.160 --> 00:20:23.119 And we were too early. 00:20:23.359 --> 00:20:36.319 I think AI has has the potential to make that capability going back to the RFP, that you know, a request for proposal is really uh you could rephrase it as a request for potential. 00:20:36.960 --> 00:20:40.480 What is the potential you can bring to me as a partner in this? 00:20:40.880 --> 00:20:54.319 You may or may not, based on your PPAT process or other requirements, depending on your industry, uh be able to change the part specifications, uh change the metal composition, a slightly different alloy, you know, change change a coding. 00:20:54.640 --> 00:20:55.519 But maybe you can. 00:20:55.759 --> 00:21:02.480 And again, it depends so much on on what it is, you know, you're you're ultimately making in terms of finished product. 00:21:02.960 --> 00:21:16.240 But I think with AI, the potential for RFPs to be that much more valuable because you know, the whole notion is I am not locking in necessarily exactly what I want to get back, right? 00:21:16.400 --> 00:21:22.319 You know, I want to bring your capability and creativity, and maybe our true north is cost savings. 00:21:22.480 --> 00:21:25.599 Maybe our true north is some other aspect, you know. 00:21:25.759 --> 00:21:27.359 So it's uh it's fascinating. 00:21:27.680 --> 00:21:28.559 It is interesting. 00:21:28.720 --> 00:21:38.319 I mean, uh, I think that that you've just sort of you've just triggered a a whole nother topic area, which is is cost savings like the primary objective of procurement? 00:21:38.559 --> 00:21:40.960 And that's a topic that's extremely hot. 00:21:41.119 --> 00:21:43.759 And I don't mean hot like it's it's well explored now. 00:21:43.839 --> 00:21:55.039 I I mean it's hot like it's nobody wants to really touch that because, you know, at the end of the day, you know, there's a lot of value that procurement can bring beyond cost savings, and people don't want to think of ourselves as, you know, just negotiators. 00:21:55.200 --> 00:21:59.759 But but I do feel that price is a universal uh medium, right? 00:22:00.319 --> 00:22:04.400 And so I can solve any problem within a business with money. 00:22:04.559 --> 00:22:10.480 And so at some level, like you have to have some kind of common currency for to describe things like performance. 00:22:10.640 --> 00:22:14.799 And it's funny because even at Tesla, we had a trade-off that we would make for the engineering team. 00:22:14.880 --> 00:22:20.240 We'd say, you know, an engineer would come to me and say, hey, I can make this part out of aluminum instead of steel, and it will be lighter. 00:22:20.400 --> 00:22:21.200 Okay, that's great. 00:22:21.279 --> 00:22:22.559 We want to reduce vehicle mass. 00:22:22.640 --> 00:22:24.880 That is like a specification that we are tracking. 00:22:25.039 --> 00:22:26.880 We had a bill of materials with cost. 00:22:27.039 --> 00:22:28.559 We also had a bill of mass, right? 00:22:28.640 --> 00:22:31.359 So we're always tracking is the car getting heavier or lighter? 00:22:31.440 --> 00:22:32.559 Now, why do you track that? 00:22:32.640 --> 00:22:35.200 Because if the car is heavier, then you're reducing the range. 00:22:35.279 --> 00:22:41.759 You might miss some other functional spec, but you could also fix the range issue by just putting more battery cells in there, right? 00:22:41.920 --> 00:22:43.519 And which makes the car more expensive. 00:22:43.680 --> 00:22:48.319 And so on the finance team, we came up with an equivalency, which was dollars per kilogram. 00:22:48.480 --> 00:22:53.440 So if you can remove one kilogram from the car, we will spend$3 extra per vehicle on that. 00:22:53.519 --> 00:22:58.160 So if you show me that it's less than$3 that you're saving by adding mass, you can't add the mass. 00:22:58.319 --> 00:23:03.440 If you're reducing mass, but you're it's costing more than$3, uh, then we we won't approve it either. 00:23:03.680 --> 00:23:09.839 Now, the funny thing is I once met my counterpart at uh at SpaceX and uh they had their own number. 00:23:09.920 -->

00:23:15.359 And you can imagine that it was like I don't remember the exact number, something like$500 per kilogram. 00:23:15.519 --> 00:23:20.000 You know, something insane like that because their cost basis is they're launching a rocket into space. 00:23:20.079 --> 00:23:25.279 And 80 to 90 percent of the cost of a rocket is the fuel to get it into space, right? 00:23:25.440 --> 00:23:31.279 Like the payload, the the amount of weight you can carry is is a fraction of the fuel weight and cost. 00:23:31.599 --> 00:23:32.640 Exactly, exactly. 00:23:32.880 --> 00:23:41.680 So that's why over there they were doing all this incredible material science research on uh carbon fibers and additive manufacturing and 3D weaving and stuff. 00:23:41.759 --> 00:23:51.279 And it was always much harder for us at Tesla to justify that because I mean, frankly, a stamping machine and uh injection mold machine, very, very efficient uh on a cost basis. 00:23:51.440 --> 00:23:55.599 But I I guess uh sorry, so circling back, you know, something I'm curious about. 00:23:55.680 --> 00:23:58.480 So I think that, okay, RFQ, we talked about that. 00:23:58.640 --> 00:24:03.359 You have a very clear specification, you're just really pushing the suppliers to compete on price. 00:24:03.599 --> 00:24:14.880 RFP indirect in the direct space, you know, we're saying, hey, we have a functional spec we need to get to, but we're opening the aperture a little bit of the things that we're willing to consider as supplier-provided solutions. 00:24:15.039 --> 00:24:25.839 Now, the drawback, of course, is that if you choose a supplier's solution, you don't have a lot of competitive pricing leverage at that point because the the point is that this is the solution that they've proposed that they've created. 00:24:26.000 --> 00:24:28.000 What about RFPs and the indirect side? 00:24:28.079 --> 00:24:34.319 You know, I'm I'm curious, you know, what do you think that the indirect world could learn from the direct materials world as someone who spent so much time there? 00:24:34.480 --> 00:24:54.000 Because a lot of times I do feel like indirect, it sort of has sometimes like a lot of the best tools in technology in the form of like the S to P intake and orchestration players, but also it's sometimes the I feel that the kinds of things people are buying are so varied that it ends up maybe not having the level of specificity that it needs. 00:24:54.079 --> 00:24:56.240 And so I'm kind of curious, what do you think? 00:24:56.319 --> 00:25:02.960 I'm I don't mean to to prime the pump too much and bias you, but what do you think are some things that the indirect world could learn from direct materials? 00:25:03.279 --> 00:25:14.400 I'll back up in terms of indirect and talk about the unstated purpose of these e-procurement solutions, going back to Ariba and through to Coupa and others today. 00:25:14.640 --> 00:25:18.640 When a company puts them in, you know, provided there's some oversight. 00:25:18.960 --> 00:25:22.880 Do you know often the first thing that happens if there hasn't been one before? 00:25:23.200 --> 00:25:24.160 No, what happens? 00:25:24.480 --> 00:25:27.200 People buy less because they're being monitored. 00:25:27.359 --> 00:25:30.240 So there's no stockpiling of toner. 00:25:30.400 --> 00:25:33.359 There's nobody upgrading a notebook every year or two. 00:25:33.680 --> 00:25:37.759 So in theory, a watch clock, you know, does behave differently. 00:25:37.920 --> 00:25:41.680 You know, we could we could go quantum mechanics around this as well. 00:25:42.160 --> 00:25:46.400 But indirect, how how much of indirect spend is really necessary? 00:25:46.640 --> 00:25:49.200 You know, the answer is probably a significant portion of it. 00:25:49.279 --> 00:25:52.000 But then we can get to over what time horizon? 00:25:52.160 --> 00:26:00.640 If the coffee maker breaks, yes, we're we're going to have a productivity loss, you know, and light sources RD team, especially at two in the morning. 00:26:00.960 --> 00:26:05.920 But, you know, at the same time, the Starbucks is around the corner or insomnia cookies. 00:26:06.240 --> 00:26:09.039 So I think I look at indirect. 00:26:09.279 --> 00:26:10.880 I'll answer the question differently. 00:26:11.119 --> 00:26:15.519 I look at indirect as a completely different animal than direct. 00:26:15.920 --> 00:26:19.039 You know, direct also is different by industry, right? 00:26:19.200 --> 00:26:25.039 To me, direct is is what is what is like the lifeblood of what you ultimately produce. 00:26:25.359 --> 00:26:27.839 It could be ingredients in the food space. 00:26:28.000 --> 00:26:31.279 It could be the importance of packaging and CPG products. 00:26:31.519 --> 00:26:45.200 You know, and certainly in manufacturing, the two of us have had a lot of experience in our years on all the areas to build a car or an airplane and everything that goes into it from raw and semi-finished materials into components. 00:26:45.519 --> 00:26:56.160 The issue with indirect is that how a business operates, what its requirements are with indirect, the factory is not going to shut down if you cut the T and E budget. 00:26:56.319 --> 00:27:00.799 You know, now is cutting is cutting the T and E budget a good idea? 00:27:01.359 --> 00:27:04.880 Shutting it off is as has happened in so many former recessions? 00:27:05.200 --> 00:27:05.759 It depends. 00:27:05.920 --> 00:27:10.160 There's probably a ton of extraneous travel, which is not necessary, but some of it is hugely strategic. 00:27:10.240 --> 00:27:15.279 You might miss a deal because your REM is not going out and socializing something early enough. 00:27:15.519 --> 00:27:25.119 So I think indirect ultimately becomes much more of a demand management game and also one of a multi-channel game as well. 00:27:25.359 --> 00:27:28.960 Not only, you know, are we looking for preferred buying channels? 00:27:29.119 --> 00:27:32.000 My old colleague Pierre Mitchell

at Spend Matters. 00:27:32.079 --> 00:27:43.039 And now hack it, back to Hackett, 25 years ago, going the phrase guided buying, you want to guide guide buyers to the best choice for them, which is in line with the company's standards. 00:27:43.279 --> 00:28:05.599 You know, direct is much more about direct engagement with the with the end supplier, whether you're beating them up on price, whether you're asking for creativity, whether you're designing a supply chain with multiple suppliers, versus indirect, which is aligning what a user really needs, also questioning if they need it with different paths to get there. 00:28:05.839 --> 00:28:18.960 So, yeah, that could be a punch out catalog, that could be a punch out catalog tied to a GPO contract and a rebate on the back end, and and that could be the complexity of corporate services and travel. 00:28:19.119 --> 00:28:21.920 It could be a bunch of things, but it's wholly a different animal. 00:28:22.160 --> 00:28:25.119 I came at indirect much later in my career at free markets. 00:28:25.200 --> 00:28:26.240 We focused on direct. 00:28:26.400 --> 00:28:28.640 You know, we would often poo-poo indirect, right? 00:28:28.799 --> 00:28:33.920 I mean, that's what Ariba, Commerce One do, and Medium and Tolesys and all those guys back in the day. 00:28:34.160 --> 00:28:40.480 But it's still a huge amount of spend, and there's you know tr tremendous potential value in going after it. 00:28:40.720 --> 00:28:41.680 Makes a ton of sense. 00:28:41.839 --> 00:28:44.079 Well, I I know that we're about at time, Jason. 00:28:44.240 --> 00:28:49.440 Is there anything that I I didn't ask that you feel that I should have asked or you would have expected me to ask? 00:28:49.839 --> 00:28:55.119 I guess there's a question in manufacturing is is do we need procurement at all? 00:28:55.279 --> 00:29:02.640 I'm focusing more in my new startup on retail and CPG and food, grocery, those areas, which is new for me. 00:29:02.720 --> 00:29:03.759 It's interesting learning. 00:29:03.920 --> 00:29:07.599 And the answer is, do you need a merchandising function in the future with AI? 00:29:07.759 --> 00:29:08.799 You probably still do. 00:29:08.960 --> 00:29:12.079 But in manufacturing, does procurement matter anymore? 00:29:12.240 --> 00:29:21.599 I think when you have a team of super design engineers and a crack supply chain team and with better data and AI in the middle, is procurement just creating friction? 00:29:21.759 --> 00:29:25.200 I don't have the answer, but I think that's the question I would have asked. 00:29:25.279 --> 00:29:26.880 And I don't know how to answer it if you did. 00:29:27.200 --> 00:29:27.440 Okay. 00:29:27.599 --> 00:29:28.319 It's interesting. 00:29:28.480 --> 00:29:34.880 I'm I mean, look, like all like all questions that are difficult to answer, probably the the truth is is the nuance, right? 00:29:34.960 --> 00:29:39.200 That in certain circumstances, uh the answer is absolutely procurement is an

obstacle. 00:29:39.279 --> 00:29:44.079 And in certain places, you know, a great procurement person is worth their literal weight in gold. 00:29:44.319 --> 00:29:46.640 And also to some degree, it depends on the outcomes, right? 00:29:46.720 --> 00:29:48.319 Like I don't want you know, it's sort of funny. 00:29:48.400 --> 00:30:07.359 Like I remember um I had a friend of mine who was at Airbnb, and the way that they they viewed the functions at Airbnb was that engineering was like the prize, prize team, and that many of the other functions around were designed to rem specifically get as much off the engineer's plate as possible. 00:30:07.519 --> 00:30:10.640 That wasn't just them coding and working on the product. 00:30:10.799 --> 00:30:19.039 And you know, it's very when you have a very you know selective, high talent, high comp group, uh, you know, sometimes it's worth enabling, surrounding them with enablement, right? 00:30:19.119 --> 00:30:20.720 And and we see this with the sales, right? 00:30:20.880 --> 00:30:26.000 You bring around sales, you know, rev revenue operations, solutions engineering. 00:30:26.400 --> 00:30:29.279 I don't want my sellers doing invoice receipts. 00:30:29.440 --> 00:30:31.599 I want that to be, you know, on accounting. 00:30:31.680 --> 00:30:43.920 And so in some ways, you might ask the question like, could a design engineer go and be a cost modeler and go and do sourcing and RFQs when it only would make up 5% of their day and they might just not be used to it. 00:30:44.079 --> 00:30:54.079 I do think there is value in specialization and certainly in the automotive arena, which is a place where we play a lot, you know, they don't they don't have teams of like five procurement folks like a lot of businesses do. 00:30:54.160 --> 00:31:01.359 We're talking about like 500, and it's not a maybe they could get away with 490, but it's it's not a mistake to have the function in its entirety, I think. 00:31:01.440 --> 00:31:01.680 You know? 00:31:02.000 --> 00:31:03.519 I want to close by asking you a question. 00:31:03.759 --> 00:31:03.920 Sure. 00:31:04.400 --> 00:31:07.839 It's not so much a procurement one, but it's one you've lived at Tesla and elsewhere. 00:31:08.000 --> 00:31:12.480 Aside from Tesla, does the US have have an automotive industry in 30 years? 00:31:12.880 --> 00:31:15.440 That is a really uh good question. 00:31:15.680 --> 00:31:25.200 I think the answer to that question, I think, maybe solved by a different question, which is that I don't believe in 30 years there will be such a thing as personal car ownership anyway. 00:31:25.359 --> 00:31:35.920 And so then I think it will end up being that there's a transition toward autonomous modes of transport that are owned by by businesses that are primarily in the arena of servicing. 00:31:36.160 --> 00:31:40.799 I sort of have this view of I I'm a big clean Christensen fan. 00:31:40.960 --> 00:31:42.400 I've re read a lot of his stuff. 00:31:42.480 --> 00:31:44.480 I went to Harvard Business School, of course. 00:31:44.559 --> 00:31:49.440 You know, he's a bit of a small uh deity there, uh, even posthumously. 00:31:49.599 --> 00:31:57.599 And there's this curve that I love, which is the levels of integration versus modularity against the state of a technology. 00:31:57.680 --> 00:32:02.799 And so right now the state of autonomy has under is underperforming where people want it to be. 00:32:02.880 --> 00:32:04.799 It's gotten a lot better, but it's not perfect. 00:32:04.960 --> 00:32:22.640 And so the solutions that really win there are the high cost, sophisticated, fully integrated end-to-end solutions where they can just get slightly better, but closer to narrowing the performance gap, then eventually it'll be that okay, autonomy is uh ubiquitous and it's not actually that special, not that interesting. 00:32:22.799 --> 00:32:30.079 And then what people will compete on are other things like can we be very efficient or can we have service delivery models that are better than others? 00:32:30.240 --> 00:32:40.799 And what that'll mean to me is that if the industry sort of breaks apart to a place where you have, you know, these service providers, maybe you have networks, like Waymo is already feeding into Uber and Lyft as a network. 00:32:40.880 --> 00:32:53.920 And then what's going to end up happening is we're gonna have a financial instrument, you and me, that we can we can buy, which will be that a hedge fund will purchase a fleet CapEx uh with a very deterministic outcome cash flow. 00:32:54.079 --> 00:33:07.359 They are the most efficient capital providers, they will securitize it across regions, or maybe not, and then you can buy the cash flows from a specific region that's kind of an equivalent to uh GDP for that city or something, you know. 00:33:07.440 --> 00:33:10.880 So I'll token I open up blockchain and on Mars. 00:33:11.119 --> 00:33:11.440 Yeah. 00:33:11.599 --> 00:33:13.440 So you're you're you're you're right. 00:33:13.519 --> 00:33:15.759 But I I don't I think 30 years is a long time. 00:33:16.000 --> 00:33:35.680 I don't think it'll be surprising to me if not only if people are a if people still buy cars, I would be surprised to be honest, that if it's even legal to drive your own car, I think it'll be like equivalent to if you're if you're in an accident and it turns out that you were drunk at the wheel, it doesn't matter if you're in in the right or the wrong, you're the one who's liable just by virtue of being drunk. 00:33:35.759 --> 00:33:36.640 I think it'll be like that. 00:33:36.720 --> 00:33:44.000 Like you can drive your own car, but if you do that and you get into an accident, it's your fault because humans are just gonna be worse than autonomous drivers. 00:33:44.079 --> 00:33:46.880 And like, look, it it happened with uh with horse and carriage. 00:33:47.039 --> 00:33:51.039 You know, there was a period of time where you could drive automobiles and ride horses on the same roads. 00:33:51.200 --> 00:33:57.920 And then now, if you try to take a horse on a an interstate freeway, uh it's it's quite damning, quite illegal. 00:33:58.000 --> 00:34:01.200 And so I think self-driven vehicles won't be around in 30 years. 00:34:01.279 --> 00:34:14.239 And but I do see like there's a lot of companies that are doing great, like like Ford in particular, and I won't talk too much about our relationship with Ford, but they have an amazing uh RD center building next generation vehicles uh in California, right? 00:34:14.320 --> 00:34:18.800 And so I think there's a lot of ex Tesla folks who have gone to these automotive companies. 00:34:18.880 --> 00:34:26.480 I think that the leadership at some of them understand that there's existential threats, and then they will find alternative ways to you know reinvent themselves. 00:34:26.559 --> 00:34:27.599 And if they don't, then they die. 00:34:27.760 --> 00:34:29.280 I mean, that's the reality of any business. 00:34:29.519 --> 00:34:30.639 Great question, Jason. 00:34:30.800 --> 00:34:32.880 I should have you be the interviewer on this podcast. 00:34:33.119 --> 00:34:34.320 Similar teas here, Joe. 00:34:34.400 --> 00:34:35.280 Thank you so much for the time. 00:34:35.360 --> 00:34:36.880 It was really great to see you. 00:34:37.119 --> 00:34:38.320 Great to see you, Spencer. 00:34:38.400 --> 00:34:39.679 Thanks for having me on. 00:34:39.920 --> 00:34:41.599 Of course, thank you for joining the pod.