Episode #56: Can AI Give Procurement a CRM Moment? Spencer Penn on Modernizing Direct Materials Sourcing
In today's episode of U.S. Manufacturing Today, Matt Horine interviews Spencer Penn, Co-founder and CEO of LightSource, an AI-powered procurement platform for direct materials. Penn recounts leading sourcing for Tesla’s Model 3 scale-up and seeing $30B of parts managed via spreadsheets and email, inspiring him to found LightSource in 2021. The conversation argues procurement is strategically critical yet undertooled compared with sales, with incentives and visibility lagging despite large P&L impact. Penn describes how AI can improve supplier discovery, should-cost estimation, negotiation support, ongoing price monitoring, and faster deployment via automated data ingestion. LightSource customers reportedly cut RFX cycle times 25–60%, improve supplier experience, and reduce cost creep; one A/B test showed 25% faster cycles and 47% lower cost creep. They discuss dual sourcing, nearshoring/onshoring realities, China’s industrial base, and advice for leaders to experiment with tools like Claude/Claude Code.
Links
- LightSource AI
- Matt Horine on LinkedIn
- Spencer Penn on LinkedIn
- Veryable Is Revitalizing U.S. Manufacturing
- Navigating Trump 2.0
- Sign Up on the Veryable Platform
Timestamps
- 00:00 Show Intro and Big Question
- 00:31 Meet Spencer and LightSource
- 02:26 Tesla Scaling Lessons
- 05:52 Elon Leadership Anecdotes
- 07:29 Why Direct Procurement Lags
- 09:06 The Real Money in Parts
- 14:01 Speed as Competitive Edge
- 16:35 Why Procurement Is Undervalued
- 22:51 AI Use Cases in Procurement
- 28:41 Fast Implementation Reality
- 30:46 LightSource Results and ROI
- 33:48 Nearshoring and China Reality
- 38:08 Free Advice for Leaders
- 42:12 How to Get Started
- 43:28 Wrap Up and Call to Action
Episode Transcript
Matt Horine: [00:00:00] Welcome to US Manufacturing Today, the podcast powered by Veryable, where we talk with the leaders, innovators, and change makers, shaping the future of American industry, along with providing regular updates on the state of manufacturing, the changing landscape policies and more.
Today, we're talking about supply chain and procurement, two of the most strategically important and historically underinvested functions in manufacturing, and we're asking a big question. Can artificial intelligence finally give procurement teams the tools they need to compete at the speed of modern manufacturing?
Our guest today is Spencer Penn, co-founder and CEO of LightSource, an AI-powered procurement platform built specifically for direct materials, the components and parts that manufacturers buy to put into their products. Spencer founded LightSource in 2021, after a career that took him from the factory floor of Tesla's model three program to product leadership at Waymo.
Spencer holds a master's of science and engineering from Harvard and an MBA from Harvard Business School, and before founding LightSource, he was one of the key leaders behind Tesla's most [00:01:00] ambitious manufacturing challenge, getting the model three to market at scale. He oversaw the sourcing of billions of dollars in direct materials and experienced firsthand what happens when procurement systems can't keep up with the pace of production.
That experience, managing 300 billion in parts on spreadsheets and email chains became the founding insight for LightSource. Today, the platform is used by hundreds of companies across four continents with customers, including major OEMs, Fortune 500 manufacturers, and leading consumer brands. LightSource has raised from investors, including Lightspeed venture partners and Bain Capital Ventures.
Today, we're gonna talk about why supply chain and procurement are finally having their technology moment, how AI is changing the way manufacturers source and manage direct materials, and what the future of supply chain looks like as reshoring tariffs and global uncertainty force companies to rethink how they buy.
Spencer, welcome to US Manufacturing Today.
Spencer Penn: Matt, thank you for having me on the program. I'm a big fan. I loved the introduction, and I wish I could have you do that for me everywhere I went.
Matt Horine: We try our best [00:02:00] here. That's one of the things that we really hope to, to nail. Hopefully nothing factually incorrect in that, but we'll get into the details.
So we are excited to have you, and I think just for our audience, we've talked before, and I think one of the most compelling parts of your story is from Tesla to founder. There's a pretty famous founder there at Tesla. Let's start at the beginning. You spent several years, including leading the sourcing strategy for the model three.
What was that experience like and what did it teach you about manufacturing at scale?
Spencer Penn: It's hard to compress the, the experience being at Tesla down to just a few short seconds, but when I first started at Tesla, it was back about 11 years ago. Tesla was a boutique car business. And I remember hearing about Tesla thinking, "Oh, that's neat.
An electric vehicle, what a concept." When I first moved to California from the East Coast, my friends and family were saying it's a big career mistake to work at Tesla, that it will never be someone's main car, that it's a tech toy or an iPad with wheels on it. But I was just excited to see what this Elon guy was, was all about.
I helped lead, as you mentioned, the Model three program, [00:03:00] which was our first mass market electric car. We scaled it from a thousand vehicles per week to 10 million over the life of the program, so really substantial growth. And I saw a million fires all over the place. So anywhere you look, there were things that are, that were broken that needed fixing.
And the thing that really spoke to me was procurement and supply chain because when I looked at the vehicle, I realized that our ability to hit our dual mandate. And our dual mandate was to deliver the car faster to market in half the time of a traditional, uh, OEM. And the second was to keep our bomb cost low such that we could sell the base vehicle for 35,000 bucks, which as you can imagine, is a lot of car for the money.
And then when I looked at the procurement team, it was like, wow, they're sourcing 30 billion dollars of car parts using spreadsheets and emails and that just drove me nuts. And so at a certain moment, I went out to the market to try to see, is there some great software for direct materials teams that we could buy?
And when you find nothing at a certain point, it's a very classic kind of founder story, but [00:04:00] you either get frustrated or inspired or some mixture of both and think, okay, I can do this better. I know what I'm looking for. Let's just go out on our own and build it. And it's been an amazing adventure and the business has scaled and we support a lot of companies.
And it's, it's been an amazing journey. Tesla in the early days though, I will say that procurement was a lot more like sales than purchasing because when you have unproven volumes, you have to convince your supply base to make an investment in you, to take you seriously. You're only making a thousand cars per week.
Today, why should we build out and plan factory lines for your production? Why should we get aggressive on pricing? You're asking us to quote you out as if you were already one of the big three automotive players. There were some suppliers that just would not give Tesla the time of day, or if they did, it was like we had to beg them for changes and they'd be, we'd be lucky to buy from them, that kind of thing.
It's bizarre actually and s- surreal for me to see the car that was unveiled and it was just a, a, a sort of [00:05:00] a prototype that we pre-sold. And then now in San Francisco, you can't go anywhere without seeing a Model three or a Model Y, which is the sort of derivative of that same platform. So it's just been incredible.
And then Elon going from ... I was always impressed by him and he was definitely the richest person I knew, but now seeing him become one of the richest people, not just on Earth, but of all time is a ... It's just an, it's unusual to see. I don't think that happens every day.
Matt Horine: No, I don't think so either. It's, I, a couple of things I lifted out of that.
One, for any of our listeners who are still doing the spreadsheet model, there's, uh ... So one, it's possible you launched the model three based off of that kind of thing, but there's a way, there's a path forward which we'll get to in the show and exciting. Any interesting anecdotes from your time with Elon kind of working and seeing that on the factory floor?
You heard the stories way back in the day of he was staying at Tesla, making that happen, and he drove that organization. Anything that worth mentioning?
Spencer Penn: Everyone has an opinion about Elon. I think that's the case today, and some folks love him and they're total fanboys. I'm not [00:06:00] in that camp. Some people hate him.
I'm also not in that camp. I think he is a really valuable asset, and I'm thrilled that he's on our side as a work, quote unquote. Elon's really brilliant. He can be a bit moody, so depending on the day that you catch him, you may want to give him the good news only. He's very sharp. He works really hard.
He's a brilliant marketer, and he never would say this, but marketing is one of his greatest strengths, absolutely. And it, it can be external marketing, it could be his own personal brand, but also internal marketing. And so you mentioned the idea of him sleeping on the factory floor. Yes, that did happen.
And when I was at Tesla, I felt like, this is so silly. He's the CEO of this multi-billion dollar company, and he has this entourage of, of assistance just completely taking a conference room and decking it out with pillows so he can stay there. But there's a hotel right across the street, like, why not just go there?
I'd pay for the $200 if that makes him more comfortable. But as I grew myself as a leader, I realized, okay, it's not about [00:07:00] him actually being in the factory. It's about the signaling to the organization that he is rolling up his sleeves, that he's on the factory floor, that he's here and he's gonna stay here until the mission is accomplished.
And I think there is no job too small, and I think great leaders often do that. But it was funny going into meetings knowing that he had slept in that room.
Matt Horine: That's... Yeah, that is a funny byproduct of being in the building when that stuff is going on more of legend now, but those are great insights and people do think about it a lot because he is a very complex businessman, is the way to, to describe that.
One of the things that you described there in, like, the opening is managing 30 billion in direct materials on Excel spreadsheets and email chains. For people who haven't been in that seat, what does that look like day-to-day and when did you have that moment
Spencer Penn: where you're like, "This doesn't exist to do it outside of this quite yet."
So just to take a step back, for folks who may not know, procurement is broken down broadly into two categories, direct materials and indirect. Direct is anything that a company buys to put into its product. So this would be [00:08:00] Tesla buying car parts. They're going into the vehicle, they're getting sold.
Tesla also bought office chairs, laptops, toilet paper. These are all procurement, but they're really more like purchasing. If you wanna think about it like that, it's things the company buys to consume itself. There's a lot of solutions focused on procurement of stuff. You can go to Amazon Business, you can use a P2P system like Coupa, punch out, you can do credit cards like Ramp or Amex.
There's many ways to buy stuff, but then on the direct material side, there's a real dearth of technology solutions and there have been, and we find that 70 plus percent of procurement teams use Excel and email as their primary tool and direct side. I like to contrast this with sales teams where 87% are using a CRM like some kind of customer 360 every day.
I think there's three realizations. One is the dollars of play are enormous. Two is that there wasn't a tool that could help solve it. And then three is that this is not just a Tesla problem, this is a mini company kind of problem. And those three things [00:09:00] taken in concert was motivation for me to, to go out of my own and really build something.
Just to give a sense of the scale of the dollars, I don't think people really think about it, but we are making 10 ... Our plan in the program is to make 10 million model threes. And I guess, Matt, I'll ask you a question like, how much do you think a steering wheel costs for a typical vehicle? If you to just guess, OEM buys a steering wheel from their tier one supplier.
It doesn't come with an airbag in it, and it does, but it does come with switches, for example, and it's leather wrapped. Do you think that's more or less than $1,000? Is it more or less than $10? What do you think it is?
Matt Horine: I'll take a guess, and let me just frame this. My background in any kind of procurement was when I was in the military.
So those price points may be a little different, and so I'll take a shot at it, but probably 800 bucks is probably what my guess would be.
Spencer Penn: Okay, 800 bucks. In the military, probably it's 800 bucks cost plus. For a reasonable OEM, a big three auto manufacturer, they should be able to get the steering wheel for, um, a premium [00:10:00] vehicle anywhere between 60 and $70 per year.
You can see how big the error bars are. With our, with our Coke Zero, I know, okay, 75 cents is cheap, a dollar is Costco price, two bucks is if I'm at, at a restaurant, eight bucks is at a stadium. And I know these things because you're buying them all the time. But when you think about the entire huge array of things that companies are buying and how heterogeneous it is, the error bars get really large.
Now, one of the other challenges that we dealt with in a lot of parts is that the cost would start at a certain price and it would creep up because when you're going into the market to negotiate with suppliers, they're competing to win the business. They're sales guys, right? So they wanna offer good pricing, optimistic pricing to win the, uh, the bid, but then once you've spent 250K qualifying the tooling and kicking off CapEx with them and vetting their factory line, you've gotten prototypes, you've iterated five more times with engineering, you got T1 parts off the initial tool and eventually you're in production, changing suppliers is very [00:11:00] expensive.
So every 5% change would become a 25 cent cost up. And again, just to reframe it in the scale of the dollars here, let's say that you're a procurement manager and you win or lose a $2 negotiation on a door handle. It seems like a drop in the bucket versus a $35,000 car, but, but there's eight door handles in the vehicle, so that's 16 bucks and then you're making 10 million cars.
So that's $160 million of EBITDA. That's $160 million of profit swing, depending on whether or not you gain or lose just $2 on a given part. And so if you're in procurement, I've had this realization that these are the most valuable people in the entire company because while they are under equipped with high quality technology, often they're under equipped with resources, time, training, et cetera, and they're not frankly appreciated that much.
If you, if you're in procurement and stuff doesn't arrive, you're in trouble, but if you do everything right and you save the company tons of money, nobody's looking [00:12:00] at you and saying, giving you a big high five, they just say, "Okay, you've done your job properly. It's like running water." But these folks can single-handedly make a difference of $100 million to the bottom line of the P&L.
So if I were a CFO, I would just be like, "What is everything in my power that I can do to en- en- enable procurement to do the best possible job that they can? " So the magnitude of dollars, then going out to the market finding, "Hey, there's no technology that really solves this. " And I still see this today. I really feel like LightSource is one of only maybe two or three solutions on earth that are really purpose-built for direct materials and teams that are building products and launching them at velocity.
And then the third thing is when I was in business school, I had a startup class and you, as part of the final project, you just iterate on an idea and I used this concept of LightSource as the idea for that project and I started interviewing folks willing to dedicate 15 minutes their time to a lowly student and I realized this is not just a Tesla problem, this is like every company across the [00:13:00] world.
If, and even if the CFO or the chief procurement officer thinks that they've implemented some kind of digital solution like an ERP system, what they'll find is that's really a, a transaction system for the actual purchase orders, but all of the battle is won or lost in the interface between the engineering system, the procurement team and the suppliers, which happens upstream.
By the time you kick off a PO and you're actually buying stuff, you've locked in your pricing. And, and in fact, 80% of the cost of a product is determined at the time of design. And in the traditional legacy industries, they use a waterfall process, if you want to use the software analogy. So engineering designs it, freeze.
Procurement sources it, freeze. Suppliers make it and manufacturing assembles it. And then you ask, why did it take us four years instead of competitors getting to market in two? Why are we 25% over our initial estimate? And the answer is that there was not strong collaboration and a, a good digital environment between these different [00:14:00] parties.
I see it all the time. I think that you never waste a good crisis. I feel like Tesla has done one thing really well that maybe folks don't talk about, which is that it is proven a new model can work, that there are these breakthrough organizations. And I would argue that companies like Anduril, which are disrupting the aerospace and defense arena, or a customer like Hims and Hers, which is moving faster than Pharma could have ever possibly imagined, or new automotive companies that are, they're beating out the old guard.
I think that Elon, Elon's real superpower is not to influence other people, it is to resist influence. And in the face of people saying, "Hey, this is not possible," he was somebody that proved that it is. And now there's an entire generation of, I would call them challenger companies that are coming up and actually winning against the legacy incumbents and the metrics that they care about are different than traditional procurement metrics.
They're not thinking about the old KPIs [00:15:00] which were reduced cost and reduced risk. Those are the two, like the trade off, cost and risk. They're winning on speed, which is a completely new metric. They're winning on velocity. They're getting the market before the legacy incumbents can even get their first prototype out the door.
And that's where LightSource really works well for those kinds of companies that are in that next generation that don't accept the world for what it is and, "Hey, this is the way, quote, we've always done it, but are trying to win and are taking a broad view of competition, not just domestically, but of folks even coming out of China, where they're doing things at a mind boggling speed, and I want us to be able to, to win and have industry here."
Matt Horine: Yeah, no, that's a critical point is that speed is the 21st century make or break for miss manufacturers, right? Like we have this great moment going on right now in the country talking a lot about reindustrialization. People are finding out what that means. It's not just the air, aerospace industry or anything like that.
It's actual like shop floor level decision making that is ... In the era of globalization, there's a [00:16:00] million reasons why American manufacturing has been held back, but now we're starting to see the tide turn. I really like the way you frame that, by the way, that CRM moment for procurement because, you know, the thinking about managing the outflow of capital in the same way that the sales side or commercial side manages the inflow, there's not a lot of technology investment.
It's been wildly unequal because people focus on pipeline. You get to ... Any manufacturing organization, they'll talk about pipeline and the commercial guys will tell you about the pipeline, but we need like hard commitment to orders, but the technology investment, do you see that as like a cultural thing or just something that is just a technology resistance or where do you see that gap?
Spencer Penn: So it's a great question. I think about this all the time, Matt, because procurement and sales are equal and opposite sides of the same coin, sales sells to procurement, procurement buys from sales. So why is there such a disparity in terms of the compensation possibility, in terms of technologies and tools?
I'll go to the big procurement conferences and a blowout year is maybe they'll have 800 or 1,000 [00:17:00] attendees. And if you go to Dreamforce that just the Salesforce conference, there's 25,000 folks, you've got the Red Hot Chili Peppers headlining. I'm a big believer that incentives drive behavior. It's one of my, one of, one of my sort of mindsets.
And I think that part of the issue is an incentives problem, which is that sales is compensated on commission. And so if you go and sell a hundred million dollar project or a 10 million dollar project, you might take home a million dollar W2, you might get a 10% commission yourself plus some kind of kickers.
You're getting so many high fives or hand is sore by the end of the day. If you're in procurement and you go and save 50 million or 100 million dollars or avoid a catastrophic failure that would have put the business a quarter behind, again, people don't say this person is the most valuable human in the room.
They look at you and say, "If you just save 50 million dollars, did that mean you left 50 million dollars on the table last year? Like why was there so much fat to be cut? Why was there so much fluff?" Or maybe, "Oh, did the market just [00:18:00] go down and so you just are riding the trend?" People don't ask that with sales.
So I, I think that folks have not been incentivized properly. I think that procurement is undercompensated and we find ourselves actually, I'll put myselves in the procurement shoes, trying to justify and calculate and explain our value. And if you spend half your time doing the work and half your time communicating to the organization why your work matters, it's a very tough position to be in and I think technology can do that, right?
If you think about your accounting software, it's keeping a log of your transaction so at the end, you don't have to pick up all the pieces. And so I want procurement to be able to just do their good work and automate the grunge work, and then also as a secondary effect of using a tool like LightSource, be credible for the rest of the organization around demonstrating value that's been provided and make that measurable and quantifiable.
And then if you can measure it, then you can compensate on it. And if you understand, "Hey, this person is 100 million dollars P&L impact better than the next person, then that person should be [00:19:00] receiving a significant bonus." And that's how you can attract and retain the top talent. I'd rather have a procurement team, an elite team of five people who are enabled with AI technology than a team of 100 who are phoning it in.
And that's a lot of our customers think.
Matt Horine: It's a great discussion because it's something that is what, from my experience, is cultural. You know, if you look at like procurement teams or folks like that, there are some kind of like mild incentivizations for saving X amount of costs, but they don't really, I think the word is like weaponize that to the effect that what we were just talking about is how do you compete and win in a market with speed?
And so companies that you mentioned, they're him, hims and hers, are you familiar with them? They push and push until they break something or come into some kind of conflict. I think we were talking about this before, where it's how do we find ways to win and it's because you're securing the supply chain and your procurement people are leading that.
The product sells, but how do you like really weaponize the model? That's my guess. I don't know if I framed that in the right way, but hopefully that makes sense.
Spencer Penn: It does make [00:20:00] sense. We have a lot of informal slogans here, as every company does, and one of them is make direct materials your company's new competitive advantage.
And so I wouldn't use the word weaponize unless we're talking actually to an aerospace and defense folks, but definitely a competitive advantage. If you can bring your product to market in half the time of your competitors, that's really valuable. And to be more precise about it, what was the impact for Tesla of doing that?
Tesla is famous for waste. I remember in the early days, we had a lot of customer complaints about, oh, this fit and finish isn't good, or this panel gap is too wide, or the car's reliability, which we bring it back to the shop, or after we sell it, there's over the air updates, because it's a software defined vehicle.
And we did a lot of things very fast, and we had to scrap a lot of parts, we did take out a lot of machinery. So it's easy to quantify the waste, but what's much harder is how do you quantify the value of getting to market three years earlier? And where that accrued [00:21:00] was to a trillion dollars of market cap that Tesla was able to capture.
So over the same period, the last 10 years, Tesla went from, I think it was 30, 40 billion dollars of market cap to 1.3 trillion. So added a trillion over that same period, Ford has basically had a growth of 0.07%, I think it is. So really actually quite, quite flat. As an organization, you think Ford is so efficient and they're so dialed in on all their process, how is it that this ragtag team of Silicon Valley folks were able to, were able to become the investor dollarlings, it's because people see the speed of innovation.
It's one of the few companies that's actually also able to compete effectively in China. Tesla has a factory in Shanghai, and one of the big questions on every quarterly earnings report is how we're the China sales. That's a remarkable reversal of positioning from where a lot of companies are today, which is we are so worried about how uncompetitive we are versus [00:22:00] Chinese vehicles that we need to beg the government to prevent them from being able to import because we would be obliterated.
We need the handicap to survive. I would much rather be fit enough to win. And I think that's the long term, the long term value that I think that great technology and AI can bring. And then you think about, okay, where does it matter? It matters in engineering, procurement, manufacturing, because that's how you deliver your product.
And if you deliver a great product, I believe if you build it, they will come. So build a great product and folks will buy it.
Matt Horine: That makes sense because there is so much of that out there where it's like the debate about like protectionism versus, wait, how are we absolutely driving this as the winner in the market because it stands on its own merit?
Most free traders or free enterprise people think like whatever the consumer or the buyer wants, you know, where you, that's where you fill the gap, not holding these things back. So it's a really nice framing of the issue. You did bring up something there that comes up often on this show and is a major component of your company.
AI in the supply chain, we have a lot of guests on, they talk about AI, there's like [00:23:00] abstract and then there's a real world where it's like actually hitting the ground and making an impact for people. It's one of the most talked about technologies and manufacturing right now. How is AI actually being used in procurement today versus how it's being talked about?
Spencer Penn: I think the biggest challenge with AI right now is that people have an idea in mind that they know it's supposed to be helpful, but they don't know exactly what problems they wanna fix yet. Every board of directors is asking their CEOs, "What's your AI strategy?" And every CEO is asking their VPs of the different functions, "What are we doing with AI?"
And then they're going to software vendors and saying, "Hey, I need a good answer for what we're doing with AI. Can you sell me something I can bring to my next board meeting?" And I find that to be a, a, a very short-term circumstance and I always would much rather focus on what are the problems that a business wants to solve, what will move the needle, not what's, you know, priority eight, nine, 10, but what's priority one, two, [00:24:00] and three, because those are the things that matter.
And then how can we bring tools and technologies and solutions that will help them solve those issues? And so when I think about AI, it bec- it starts to sound a little bit like word mush at a certain point. And I do think that in the next five to 10 years, we're stop- we're not gonna hear the word AI anymore, we're just gonna call it software in the same that now it's assumed that any SaaS platform is on cloud, it's not on- prem.
People used to say we are a cloud-native SaaS platform. Now you just say we're a software that we're a CRM or we're, uh, in the s- supply chain side, we're an SRM, a supplier relationship management tool. I think it will become resipsa or fully self-evident that every software platform will involve AI, both in terms of the infrastructure that they build, agents in terms of the systems of not record, but systems of action.
And so if you think about where AI can actually apply within the [00:25:00] procurement landscape today, it's all the different pain points that people struggle with. So in the design side, it's working with engineering to help them modify their designs to make it more commercially, commercially savvy. Then bringing those designs to the market, who are the right suppliers?
We've got our list of three or four suppliers we've always worked with. Who are the 15 others that we might not know about? It's a lot of work to find them, but AI is able to scrape the web, pull together lots of different data sources. It's really good at certain kinds of what I call fuzzy matching. So if you search for a specific term, these models are robust.
And so it can find like terms implicitly within the model that don't have to be exact, so you don't filter for the exact term, but it can actually search in a much more broad-based way and consider other factors as well. When you bring the part to market, what's the should cost? Before we even, before we even bring the steering wheel to the suppliers, our team should be able to tell you from light source, how much does [00:26:00] this thing cost based on all of your historical data?
And then when you're actually running your negotiation, bringing the best category, procurement expertise to bear to get the right results. And the right results, by the way, don't always mean just the cheapest price. It might be the right optimization of cost, quality, on- time delivery, and strategic considerations as well.
And then even post-award, the world doesn't stop. This is where I think indirect starts to look really different from direct. And indirect, if I'm buying some reams of paper, I buy it, I bid it out maybe to three suppliers, and I'm done, and then it's at our office, I never think about it again. But if I'm buying a steering wheel, I always have to be looking at every week, every month, the price of leather hide, because that is going to factor into the cost equation for the steering wheel, or the price of steel, or die cast aluminum, or wire harness, or labor down in Juarez, Mexico.
And so there's a lot of factors to consider because when you're buying things in the manufacturing setting, [00:27:00] it's not often that you're just buying them one time. What you're really doing is buying and establishing a strategic relationship with a supply network that you have to keep a pulse on a go forward basis.
And I think part of the challenge about procurement today is that people are always fighting fires because it's very reactive. Problems surface for you down the line and you have to fix them and it's a pound of the cure instead of an ounce of prevention. And I think with AI, we only have one attention span, but if you have an army of AI interns, one AI agent focused on each of your suppliers on each of your shipments, on each of your parts, they can escalate to you some of the issues that maybe you don't have the time or headspace to think about.
So I do think that there is a lot of different places within procurement and supply chain where you can find, find value. But there's even really basic stuff, and I know this is gonna sound boring, but it does matter. We use AI for data ingestion, and one of the big issues, like practical issues in the [00:28:00] enterprise is people say, "We would love to do this, but our data is a mess, or it's in all these different locations, or our systems are hard to talk to or hard to integrate with.
You know what? My little brother is able to use Claude Code today to build out an application in a weekend. We can also use Agentic tools to understand, explore your data and connect it with LightSource in a way that doesn't require a six to six, six month, a frankly, six-year migration integration conversation with a third-party systems integrator.
You can get up and running in a week if you so desire. And that's the, that's some of the power of, of AI in terms of the actual deployment and implementation side of the house. Now, the funny thing, and I'll tell you, Matt, sometimes when I go to companies and they ask, "What's your standard implementation timeline?"
I have learned that they don't believe us. If I say a timeline that's too low, I remember we had this conversation with a large tractor manufacturer, I won't say which, [00:29:00] and their head of digitization asked us for a timeline and a rollout plan for one division, one region, and then globally, and we gave a timeline that was actually accurate and had married up with one of the big three auto manufacturers we had supported in the previous year.
And they came back and said, "I have to give you the feedback that you lost a lot of credibility when you gave those timelines because we would expect this to be a multi-year rollout." And I was, "Man, like, we're definitely not gonna be the long pull People in the tent then, but no, we've had other customers that can get up and running same week.
And when folks talk about digital transformation, they imagine a legacy tool that requires a lot of customization. They imagine a big bill from Deloitte or Accenture who, again, is compensated on hours. And so if you compensate someone on hours, what happens to the number of hours? It goes up. And at the end of the day, they get something that was more trouble than it's worth and it took forever and it was really expensive.
And it's hard to imagine a counterfactual where it's, okay, [00:30:00] we can meet this company, buy the software, implement it with their team without any third party involved, and start seeing results in the first week, and prove the ROI completely in the first month. It's just a new way that people have to be willing to think, but it's different.
Matt Horine: Yeah, that's a really interesting component of it because we were talking a little earlier about the culture around procurement and like some of these teams, right? And th- they stand on a little bit of credibility when they say, "You may have lost your credibility because of how fast you said you could do that.
That might be like a strong suit for them, but it's a way to surprise is delight the customer, I would assume, based on the outcomes that you described there. Uh, just a follow-up question on that. What does a typical before and after look like for a manufacturer that implements LightSource? What do you see usually that changes first?
Spencer Penn: The thing that changes first, I would say, is the cycle time on a given RFX event. RFX means request for X. It could be request for quote, request for information. What they'll find is that they're able to reduce their RFX time [00:31:00] by 25 to 60%. It just depends on their process. So there is wide distribution there depending on the customer.
The second thing that they'll get typically is positive feedback from the supplier. And this is really important to us. Even though the supplier is not a paying customer of ours, I think of them as a, at a core user. If the suppliers are happy, the procurement team will be happy. The suppliers will respond faster with less complaint.
You can't, you can only make a first impression once, and so we need the supplier experience to be absolutely dialed in. That's why our team, even from the very early days of LightSource, has ma- had to make sure that we are available through the ISPs in every single country across the world, except for the ones that are sanctioned.
But if you have suppliers in China, they get redirected to not lightsource.ai, but lightsource.cn. And we have a China local version of our site with an internet cloud provider license from the government of Beijing that allows us to actually serve our website [00:32:00] legally in China. So it makes it possible for your suppliers not to have to send you WhatsApps or WeChats or emails in order to interface with your business.
Supplier experience is number two. And then the third thing, and this one is, for me, the most potent, but it takes the most time, is you start to see a reduction in cost. But often what it comes in the form of is not taking cost out of the program, but for new product introduction, it is a, a reduction of the cost creep that naturally happens from the time of design through time of production.
And so I'll give an example. We had a big three automotive customer that started using LightSource halfway through their program development. And at first I was peeved because I was like, all right, I wanted the whole vehicle on LightSource, not half. But we got accidental blessing, which was an AB test.
So we got to see the parts of the vehicle that were sourced outside of LightSource and the parts that were sourced through LightSource. And with our champion there, we did an [00:33:00] analysis, and we found two big differences between those components. Same vehicle, same team, same categories, but the parts that were sourced through LightSource, the sourcing cycles were 25% faster, and then the cost creep was reduced by 47%, so roughly by half.
So if you would have seen your bomb costs go up from 100 to $150, you saw that 50 was reduced to just 25. And so we're able to see much more significant cost creep in the items that were sourced outside of LightSource, and then the prices were updated in a much more ad hoc way versus the, the continuous updates happening through LightSource.
So from a cost containment perspective, it depends on the scale of the business, but it can be, it can be a billion dollar value proposition if done properly.
Matt Horine: Wow. It's a pretty big value prop for sure. I think, you know, one thing because you mentioned the word China, there's a lot going on and taking a pivot here.
Concepts like nearshoring, onshoring, dual sourcing have gone from theoretical to urgent. [00:34:00] How does LightSource help manufacturing, or manufacturing in general execute on those strategies?
Spencer Penn: This is, this requires more time than we have, but I would say that if you want a dual source, you are by definition doubling your workload, right?
So you're saying, "Hey, I have 100 parts that I have to source to 10 suppliers, now I wanna have 20 supplier options." And so you have to double your team's work in terms of the RFX that they're doing, the request for quotes, the supplier qualification and validation from the quality teams. You're doubling your CapEx and your tooling because now you don't just have the tool for your part in one location you have it in two.
So if you wanna dual source without doubling your team, the only way to do it is can you get more out of the same organization, which means you have to enable them with technology that automates their work. And so that's what LightSource can help do. It makes dual sourcing approachable from an organizational perspective, although it still requires, of course, like, additional capital investment.
I would [00:35:00] say in terms of onshoring and nearshoring, supplier discovery is something that we can also help with that I think can be quite difficult for organizations. It's an area of active development for us. I don't think anyone has fully cracked the code on perfect global supplier capabilities and information.
We are hoping that we will be the first to really get a supplier discovery product to a place where it is, is fulfilling the promise, but it is a long journey. And I think in terms of nearshoring and onshoring, there is like one ... Sometimes these problems are too big for us as, as individual commercial actors.
I do think that there is a reality that is very tough for people to swallow, which is that there are some things that you cannot get outside of China today. And even if you would like to find an alternative source in the United States, it's not about your desire or willingness to pay. It's about the entire industrial base that's been built up there for especially electronic components, as well as [00:36:00] expertise in many other electromechanical elements.
Just to, to put it into context, you remember the CHIPS Act, the, the Biden Chips Act? I was happy about that. It was a, a $50 billion investment in American fabrication fab facilities over five years. The Secretary of Commerce at the time said it was the sing- called it the single biggest in- investment in American manufacturing premisey or something in the last decades or whatever.
There's a lot of quotes. People were very excited about it. Now, just to give you the scale, Apple was investing $54 billion per year, even back in 2015, when they were setting up Foxconn and the, all of the sort of tiers of manufacturers below them. It wasn't out of desire to, like, necessarily just improve that country's industrial base, it was because they found a very capable manufacturing partner that was willing to put policies and [00:37:00] urgency and clear the path to enable them to scale production reliably to a billion something iPhones.
And so if you find yourself doing five times the amount of investment as what is the US government's biggest industrial program in the last century per year, and you do that for 20 years, you will have a huge supply base that's difficult to replicate overnight in the United States. And so I think there is an element of, it'll take time to build up those capabilities here or outside of China.
And I think that's just an important reality to look at and understand the reasons why. And I think commercial actors, free market has been a big part of that, but also government policy on both directions. We have to make it easy to do business. We have to be reliable for folks. We have to allow them to make long-term investments and then ensure that those investments do pay off.
Matt Horine: No, that's a recurring theme is that industrial policy has to align. And I don't think ... I, you framed it [00:38:00] really nicely, by the way. People don't understand the scope and the scale of what has been invested even by American companies into China and how hard that is to catch up. This is the final segment of the show and something that I call like the tax, the free advice segment.
So, you know, from your perspective, if you're, you know, if there's a VP of operations or supply chain leader listening to this today, what's the one thing they should do differently when they get back to the office tomorrow to start getting this process rolling? We'll get specifically into LightSource, but what's something they can do different or think about differently?
Spencer Penn: I'll give my answer, but then I'm curious, Matt, what other answers other people have given in the past? My answer is something that you might not expect. I think that when they not get into the office, but when they get home from the office on Friday afternoon, I think they should get a subscription to Claude and Claude Code, and they should try to build themselves an app, try to build something for themselves, because it will be a little clunky to start, especially if you've never coded before.
But I do feel that, at least here in [00:39:00] Silicon Valley, I see the pace at which AI is moving and it's incredible. And I think that if leaders don't really understand tangibly the art of the possible today, it is gonna be very hard to actually credibly organize your teams and process and organization around this technology shift, which to me is massive.
People talk a lot about agents, they talk a lot about LLMs, they don't know what those things do. They don't feel the ... There's an element of just learning how to use these tools where you start to get a sense of where they go right, where they go wrong, and you develop an actual intuition. And I think it's actually a skill.
I think it takes practice. Like, I've built a lot of applications recently using AI that I've thrown away, and I don't regret a single moment that I've spent on those things because it allows me to be more prescriptive with my team about how they should be using AI, not just in our product and in our sales process, but actually in our own work, how can we be AI maximalists, but also do it in a way that's rational.
And you see a lot of business leaders, even Mark [00:40:00] Zuckerberg, he, like, took a week off of ... I don't know if this is true or just marketing, he took time, he's co- quote, coding again. So Mark Zuckerberg, the CEO of Meta, Facebook, tens of thousands of folks in the organization, and he's spending time to go look at how to, how to code using these Agentic AI systems.
And then once you do it yourself, you get a feel for it. I think you can really inform your teams. So that's where I would begin, but I'm curious ... And then, of course, my, my number one would be by LightSource. But Matt, I know we have limited time, but what are the best answers that you've heard to this question?
I'm curious which ones have really stood out to you.
Matt Horine: Yeah, though, that was one of them, because the answer, like you said, is often right in front of you, and what can you do today to change your operating environment? And AI has been a fantastic resource for people who aren't scared of it and more in tune with what can I build with it as an individual, an individual augmentation, and then build it broader into the organization.
The, the best answers are something very similar to what you just provided minus the AI. It's usually [00:41:00] something to the effect of go out and understand your processes. And we had a guest on very recently that said, there's a lot of guys in the, not just the C-suite, but like middle level management that have not been on the shop floor.
They have not seen how things are going. They are doing things in a way that is just the system-oriented approach. And what it does is entrenches some of the behaviors we talked about, right? Like where, you know, you don't go down and figure out what it is that your procurement people are incentivized by.
And you've obviously been a great champion for the procurement department at large, I guess we could call it after our chat today, but not understanding what it is that they do, what incentivizes them, and how they can make those changes without some kind of broad initiative or, you know, something that takes buying a software suite or having some major implementation.
So it's about people getting back in touch with their organizations has been the consistent theme, but I think I like the AI sprinkle on top there that says you can do things, you can just do things. I think that's one of my favorite answers is just do things. Do, you know, [00:42:00] what you, whatever it is that's holding you or your manufacturing process back, just go change it.
And I think people get caught up on the bureaucracy. AI has flattened that, obviously. And I think, you know, what you've described, obviously, the second part of your answer is by LightSource. If a manufacturer today is ready to modernize their sourcing process, how do they start, where do they start, and how would you recommend getting started with LightSource?
Spencer Penn: Yeah, that's a great question. So they can just send a check directly to our office, and then depending on the amount, we will give you a call. No, I think it's very simple. Just go to lightsource.ai. That's our website. It's our most up-to-date source of information. You can also see educational information about the product itself.
A lot of people will self-educate for weeks with their teams before they even contact us, and so I really encourage that. People are ... The buying journey is becoming a lot less sales driven. It's becoming a lot more self-education driven. This is another sort of micro theme we're seeing with AI as well.
So talk to Claude or ChatGPT and ask them about us. And then the second is you can just [00:43:00] reach out to me on LinkedIn. I am a very approachable, honest person, maybe to a fault, as you can tell. I, I'm not varnishing any of my opinions, and you can reach out to me on LinkedIn or at Spencer@lightsource.ai, and I'm always happy to respond.
Matt Horine: Excellent. You know what? I will take you up on your offer, maybe grabbing a beer at some point or Coke Zero somewhere in person, but this has been a great discussion and really enlightening for our audience. So thanks for coming on today, Spencer.
Spencer Penn: Touch is all mine, Matt. Thank you for having me.
Matt Horine: Manufacturing competitiveness is built on the ability to source the right parts from the right suppliers at the right cost and to do it faster than your competition. For too long, the procurement function has been the quiet center of manufacturing operations, critical to every product that rolls off the line, but underserved by the tools available to it. Artificial intelligence is changing that and the manufacturers who move first will have a genuine advantage.
To stay ahead of the curve and to help plan your strategy, please check out our [00:26:00] website at www.veryableops.com and under the resources section titled Trump 2.0, where you can see the framework around upcoming policies and how it will impact you and your business. If you're on socials, give us a follow on LinkedIn, X, formerly Twitter, and Instagram. And if you're enjoying the podcast, please feel free to follow the show on Apple Podcasts, Spotify, or YouTube, and leave us a rating and don't forget to subscribe. Thank you again for joining us and learning more about how you can make your way.