AI Supply Chain Solutions: From BOM Scrubs to Lead Time Automation

Join us for an in-depth conversation with Keith Hartley, CEO and Board Member of LevaData, as he reveals how AI is revolutionizing direct material sourcing and supply chain management. Discover how LevaData's innovative platform helps companies break free from the "decision abyss" - the gap between fragmented data systems and smart procurement decisions.

Keith shares fascinating insights on autonomous supply chains, the shift from spreadsheet-driven processes to AI-powered sourcing, and how companies can achieve 6-10% annual savings while de-risking their supply chains. Learn about the critical role of direct material sourcing in everything from smartphones to automobiles, and why the COVID pandemic became the greatest accelerator for supply chain innovation.

Resources from this Episode:
- Connect with Keith on LinkedIn: https://www.linkedin.com/in/keithhartley/
- Check out Keith's Substack: https://substack.com/@keithhartley
- Learn more about LevaData: https://www.levadata.com/

Listen to the Episode

Watch the Episode

Transcript

James: Hi everyone, this is James from the Control+Listen podcast, brought to you by Octopart. Today we have a special guest for you. This is Keith Hartley. He’s the CEO and a board member of LevaData. Thank you so much for coming on the show. It’s great to have you.

Keith: Great to be here. Thanks for having me.

James: Anytime. Just to start with, do you want to introduce LevaData, tell people a little bit about what the company does and its background?

Keith: Sure. At LevaData, we help direct material sourcing professionals. We help people buy smarter, buy cheaper, and de-risk their supply chain across their parts, metals, and ingredients.

The company was founded by an ex–Cisco Systems employee who ran direct material sourcing and grew frustrated with the tools the company had, which involved a mountain of spreadsheets, data sources, and databases. The company is really about helping an underserved class of professionals—direct material sourcing professionals—who are at the tip of the spear for de-risking supply chains, saving money, and hitting cost of goods sold and all the finance metrics people are held accountable for. Direct material sourcing is at the front of that.

The company has been around for a number of years now. We’ve acquired customers and clients and have long-standing relationships in the sourcing industry.

James: Fantastic. And for you personally, what’s your background?

Keith: My background is in all things supply chain and procurement. I like to say that I’m a supply chain nerd, and I guess I’ve become a procurement nerd as well.

I’ve had a variety of strategy and execution go-to-market roles with companies like IBM. I’ve been at Oracle running sales strategy, at a company like Ivalua, and most recently LevaData. I’ve had a career arc that has led me into really challenging things and really difficult behavioral changes for companies.

The technology is always one aspect. We can talk all day about technology—I love technology, I’m a technologist—but at the end of the day, it’s about how people at companies can change their behavior and their long-established, often entrenched patterns. What you’re really trying to change is human behavior through the use of technology.

I’ve been fortunate enough to be in a lot of different companies, in a variety of roles, that have always put me in those areas where new technology and innovative ways of changing patterns and changing human behavior are meeting people and their companies where they’re at.

James: Fascinating. I love that there’s a dual side to that, that the two go hand in hand.

So, just to break down a key point you mentioned earlier, do you want to explain what direct material sourcing is, for anyone who isn’t familiar with what that means?

Keith: Sure. If you think about products—any product you buy—there are parts, metals, ingredients, and other attributes that go into that product. Direct material sourcing is the purchasing of all those individual components or the componentry.

An example would be a cell phone. There are approximately 600 different parts in a cell phone. Whatever cell phone manufacturer you buy from, they have people who are in direct material sourcing. They source the glass, the microelectronics, the resistors, transistors, MOSFETs, diodes—things like that that go into a device.

It’s not just phones, which everyone understands. It’s your washer and dryer, your dishwasher, and of course an automobile—lots of sourcing of finished goods there. Direct material sourcing focuses on the actual components and pieces that go into a product.

When you break a product down into its bill of materials—the total list of everything that goes into a product—it’s the people at a company who source those components. A good example is Nike. Someone at Nike is concerned about buying rubber for soles, cotton for shoelaces, and all the other materials that go into the shoe. Those people are direct material sourcing professionals.

In the LevaData world, we think about people who are buying anything electronic—electronic componentry—and then also plastic parts and plastic resins, which we call custom parts. Outside of electronics, which are usually the largest percentage of the cost of a product, the next largest cost percentage is plastics, custom parts.

We work with people who care about buying those components and who worry about where market trends are going and how to de-risk their supply chain.

James: Fascinating. It’s interesting because plastics are one of those things people don’t even think about. It’s such an integral part of everything, but people think, “Oh, it’s just part of it.” They don’t think of that as a separate component of something—it’s just a given that there’ll be plastic.

Keith: Yeah, and someone, somewhere, has to buy the resins. Someone has to get a machine rate or a molder rate.

Thirty or forty years ago, when you talked about manufacturing and making products, people would buy their components, do their sourcing, and then manufacture their own product. We’re living in an incredible age of manufacturing now, where much of that manufacturing is being outsourced through contract manufacturing.

If you and I were to start a T-shirt company, we’d have a great idea to make white T-shirts with a logo. Instead of manufacturing our own T-shirts, we’d pay someone else to make the T-shirt and pay them for it—contract manufacturing.

Most of the Fortune 500 manufacturers that make products—non-banks that actually make products—are on a pendulum shift from owning their own manufacturing to contract manufacturing. Many of them are somewhere in the traversal. They’re moving more towards contract manufacturing. It makes good business sense when you acquire properties, when you move into different regions. If you’re moving from Asia into North Africa, you might start with a contract manufacturing model.

What that’s driven is a real need for intelligence on what you’re buying and real direct material sourcing competency. No longer can you just buy it and manufacture it yourself. You might now be buying it and passing it through to a contract manufacturer. Maybe you’re telling them where to buy it, what to buy, how much to buy, and at what price they should buy it. There are different business terms and language around that channel.

It’s a really innovative evolution in sourcing, and it’s evolved to keep up with the recent trend toward contract manufacturing.

James: Makes sense.

There’s another term you mentioned last time we spoke that I wanted to have you expand on a little bit: autonomous supply chain. What exactly is that, and what’s the benefit of operating in that space?

Keith: If you ask ten people, you’ll get ten different definitions.

An autonomous supply chain, if I break it down, is supply chain processes using more advanced technology. You can begin and end that wherever you want. Is it AI? Is it generative AI? Is it agentic AI? Is it machine learning? Is it robotics? Is it Internet of Things? It’s all of those things.

When you think about supply chain, people who care about supply chains are people who make products—full stop. If you make a product, you have some supply chain. Even if you’re doing a lot in Excel, even if you’re a one-person manufacturer of T-shirts, you have a supply chain. You’re buying stuff, turning it into a good, and selling it.

As companies and organizations get bigger, you have disparate pockets of supply chain processes. You have supply chain planning—S&OP, SN, retail merchandising workflows. Then you have execution: warehousing, transportation, labor management. You have those workflows. You have digital commerce, order orchestration, order validation. And then you have procurement processes for direct materials: source-to-contract, procure-to-pay.

All the value in that workflow is held up in the “S” part, in the sourcing part. Everything else is transactional and administrative. Nobody has the best AP automation system on the planet and gets competitive differentiation from that. You can either source better than your competitors, or you can’t. It’s pretty straightforward.

The autonomous supply chain means taking all these disparate workflows and truly orchestrating them to work in concert with each other, where the system is self-governing in some areas and processes like planning, ordering, scheduling, logistics, and delivery are handled by algorithms.

Today we’d say that’s moving towards agents or agentic AI. There’s AI, there’s generative AI, there’s agentic AI—all of that is wrapped up in an autonomous supply chain. At a base level, it’s taking all these disparate workflows—which get very complex as you get larger—and making them work together without having to be constantly monitored by a human.

It’s very aspirational. I don’t want to say it’s Star Trek, but it depends on definitions and where you are on the pendulum.

The exciting thing, for someone who’s been in supply chain for almost 30 years, is that there’s a recognition that software vendors—any one vendor—can’t do it alone. With the agentic world coming, you’re seeing vendors lose some control and the control shifting to the user, who can now orchestrate their own supply workflows as they like. It’s an exciting time to be in supply chain and for autonomous supply chains.

James: So I want to ask you something, and I might start with the AI section since we’re already on that topic. How crucial would you say AI is to what you do as a company?

Keith: Massive.

A little over two years ago, on kind of a lark, we applied for an AI awards program run by an organization called AI Authority. We applied, and we won “Best AI in All of Procurement,” a little over two years ago—which I will tell you I thought was strange, because I thought we were very weak in AI. It turns out we were further ahead than anybody at that time. We were just doing some advanced statistical forward variance, some advanced mathematical concepts. Apparently that was AI and continues to be AI.

We’ve been doing this for years. At LevaData, what we do is try to take all of the world’s information on parts and ingredients and blend it. We use AI to be the best blender we can be—partnering with Octopart, partnering with other partners, partnering with our customers.

I want to know every bit of data—what I call attributal data—and make it specific to a manufacturer part number so we can serve up insights. I want every data source on the planet contextualized around everything we can know about a bottom-of-the-supply-chain manufacturer part number (MPN).

AI for us is crucial internally because we’re big data geeks. We take all the data and we blend it. As a big data geek, you have to use AI to get around and to form this large sourcing model that we have at LevaData.

That’s all internal, geeky stuff, which I care about, but our customers don’t really care about that. They just want to know their insights.

We’ve now released an external, user-facing AI engine where you can ask natural-language queries and get insights and actions right there. You no longer have to come into LevaData and take the information as I give it to you. If you can construct a sentence—or even a few words—you can work with LevaData, and we can serve up insights for you.

So how important is AI to us? It’s important in driving ease of use, and it’s important in how we contextualize all the world’s data around parts and ingredients. I believe the company that knows the most about data tends to do pretty well in a market segment. That’s where we’ve positioned LevaData: we are actively grabbing every data source on the earth that has to do with MPNs and putting it into our data blender.

AI is really critical. And I still can’t believe we won the award a couple of years ago. I didn’t think we were doing AI, but they say we were—so there you go.

James: Ahead of your time. Congratulations.

You’ve covered this in the answer you just gave, but to make it a little clearer for anyone who didn’t pick up on that: what would you say are your key offerings as a company, as far as product goes?

Keith: We are software as a service. We have one product offering.

You load your spend—as a sourcing professional, you load your spend: MPNs, suppliers, quantities, when you want to buy, and price. Then we blend that with over a billion data points that we’ve contextualized, and we serve up insights.

The first insight is just seeing your spend: by quarter, by category, by person, by supplier. Then we offer savings opportunities. Often we’ll see customers who are paying more than they should and want to save money—of course, everyone wants to save money. When you come to LevaData, you come because you want to save money.

We show savings opportunities—that’s the variance between what people in the market are paying, the variance in distributor pricing, and what you’re paying. There are a variety of savings opportunities we show, but then people stay with us for all the other reasons we constantly serve up.

We also ingest BOMs—new bills of materials. We have some BOM compare capability. And then we have a full-featured RFX engine.

Once you are in LevaData as a sourcing professional, you want to take action: get new prices, recheck something, recalibrate something. We have everything summed up to the supplier and contract manufacturer level so that when you do a quarterly business review or your twice-annual negotiation, you can use the LevaData screen and have all the information at your fingertips. You don’t negotiate one part at a time; you negotiate a basket of parts. You need to know how your discounts are blended and how you can achieve 2, 3, 6, 8% cost savings while de-risking your supply chains, because we have all the risk attributes contextualized as well.

When I think about risk, I think very specifically at the MPN level: what are the lead times, what are the years to end of life, what is the age, and so on. When engineers come in and sourcing professionals get a new project—they’re doing an NPI for some new version of a product—and 30% of the parts are end-of-life, that’s not a good place to be.

You need alternate part matching, which we provide. We’ve mapped the world of form-fit-function parts across all suppliers. We know that this 50-ohm resistor and that 50-ohm resistor are correlated, and we’re able to make those matches.

We are a one-product company that solves both for insights and for workflow, because at the end of the day, I want to be the place—the toolset—that a sourcing professional lives in. Their job, in sourcing, is to go negotiate with suppliers and contract manufacturers. That is their job. It’s not to be a data analyst. Their job is to go negotiate and get a better deal for their company: whether they’re buying more or less, whether it’s new product introduction or legacy product, whether there’s a war in Israel and they’re trying to shift to Thailand, whether they’re trying to minimize exposure to China or tariffs in Japan.

They’re always chasing some use case to solve, and without fully contextualized data, they can’t chase it and accomplish it. Our belief is that we’ve created the system that puts the direct material sourcing professional at the center:

• What do you do during the day?
• I want to find savings opportunities.
• I want to de-risk parts.
• I’m going to a supplier negotiation.
• I have to talk to a contract manufacturer. What’s my point of view?

All of these workflows we solve for in the LevaData application, and that’s what makes us so powerful in the market—we think purely about the job of being a direct material sourcing agent.

James: Right, and I imagine that makes you quite scalable as well, considering you offer different solutions, and a company could find so many different uses for your platform.

Keith: Super scalable.

We have customers that have 5, 6, 7 billion of annual spend loaded into our platform, with no signs of slowing down. We have customers who load 50 million of spend. We run the gamut.

We have customers who started with one or two sourcing professionals. One very large Japanese conglomerate with tens of billions of dollars in revenue now has 45 users of our system. We started with two, we’ve gone to 45—sky’s the limit.

We have no scalability problems. I truly believe that he or she who has the biggest blended dataset wins over time. I’ve focused LevaData’s sizable brain trust on being a fast, rapid data blender.

Here’s an example: new prospects that come to LevaData, when we show them the software, always say the same thing: “Wow, that looks great. I don’t know if it works for me because my data is in bad shape. You have no idea. I’ve got a hundred spreadsheets, lots of pivot tables. It’s not good.”

We say, “Try us. Send us everything you’ve got.” Seven days later, I’ll tell you everything I know about your data. We can go live in seven days.

All of our IP around AI is in understanding character strings and datasets that are unknown. It’s exciting when we get data and our engines have to sort through it and understand how to correlate it. Then we work with the client on what their data story tells them, because again, their job is to go negotiate with suppliers and contract manufacturers, not to be a data nerd. We take care of that for them.

James: Right. I was wondering if it would be possible for you to give us a use case or two—ways that customers have really found benefits specifically with the software.

Keith: It’s always funny. Use cases are funny because every company talks about a use case like it’s the most original thing ever, and I just see the same use cases.

People start by wanting to save money. “I’m bringing a new product to market. I want to make sure I’m not getting ripped off.” Or, “We’re shifting from our own manufacturing to contract manufacturing, and we lose visibility of buying components. We need to trust but verify what our contract manufacturer is saying. I need to make sure I’m not being taken advantage of.” It’s always to save money.

When you can show hard savings—our customers tell us it’s between 6 and 10% a year, every year—and we’ve had customers for eight or nine years now, we save them money over and over because our engine is always on, showing them opportunities.

But they stay for other use cases. Some companies really care about risk. Do we have end-of-life components? Fast-moving end-of-life components can stall production. It’s a killer in manufacturing. Some don’t care about that as much.

Some clients care deeply about tariffs at the individual part level and have no visibility into tariffs. Often we’ll show them in our software: “Here are the parts you buy from China. Here is a subset of those parts that you could also buy in Japan, and at current tariff rates here’s what you can save.” It’s an incredibly powerful tool when you know locations. That’s another common use case.

I’d say two more use cases have to do with actions out of our system. One is BOM management. As a sourcing professional, you’re always dealing with bills of materials. You’re always dealing with new product introduction. Maybe you’re going from the iPhone 15 to the iPhone 16. How much part reuse? How much on-hand inventory? That whole workflow we call BOM management or BOM awareness.

We’re able to ingest BOMs, compare BOMs, and simulate: what if I changed these parts, bought them in these countries—what does the cost do? With form-fit-function parts, we can do all that in our BOM tools. That’s a very common use case.

The last one is everything RFP, RFI, RFQ-related. When you’re in sourcing and you spot an opportunity where you need to get a better price, today most people flip from their spreadsheet to Outlook, send an email, and attach a spreadsheet. If that person resigns or is suddenly gone, that trail and history is lost.

As a supply chain and procurement person, I want that as a workflow in a system where I can see, monitor, and track independent of who is doing the RFX. We built an entire workflow engine inside our application.

Most companies we deal with have a big RFP tool for indirect and direct spend—buying laptops, pencils, and so on. Our tool is built with the sourcing professional in mind, so you can communicate and collaborate directly with your suppliers and contract manufacturers.

When you have the information, you should be able to collaborate right there. You don’t have to flip over to Outlook or Gmail. You do it in the system, next to the data.

Those are some common use cases. But the bottom-line use case that 100% of our customers care about: everyone wants to save money. Who doesn’t want a better deal, if they think they can get one, or they’re open to someone pointing out, “This is how you get a better deal”?

James: Just to change direction slightly, I saw on your site that you have some thought leadership content. There were two areas that stood out to me. I wonder if you’d go into them a little: “decision abyss” and “data gulch.” Those are interesting terms. Can you tell us what those mean?

Keith: This is really exciting. I created the concept of the decision abyss, I guess maybe last year.

This is where sourcing teams, product teams, and supply chain teams are stuck between fragmented, siloed systems, processes, and communication patterns. I would often see an engineering team come to the table with some data and a point of view. The supply chain planning team would come with their data and point of view. The direct material sourcing and supply planning team would come with their data and point of view.

They were all right based on their view, but none of them had a broad enough view to understand all these fragmented systems and departmental silos to make the best decision. It creates this phenomenon I’ve called the decision abyss.

Because of how procurement, supply chain, finance, and product are acting, you get this decision abyss where they’re gripped by the fear of making a bad decision. They know they have to make a decision, but there’s this abyss in why they can’t get to the right decision, despite smart people doing smart things with the information they have.

When I first talked about the decision abyss, it was a metaphor for this lack of contextualized data. You’ve got all these data pockets, and people are genuinely trying to do their best, but they can’t see the forest for the trees.

The decision abyss is the result—the tip of the iceberg. The part underneath the waterline, the bigger part, is the data gulch. It comes from these pockets of siloed data that prevent them from taking advantage of contextualized data and the resulting effect of making better decisions that take into account all variables.

Part of the issue is that sourcing professionals are very clever. They’re very resourceful. We’ve normalized the pain of this phenomenon. As supply chain and procurement people, we normalize the pain: “I’ve been using my spreadsheet for 15 years. It’s terrible, but nobody else can help me.” That’s literally the attitude. It’s the adage: “You’ll pry the spreadsheet from my cold dead hands. This is my spreadsheet, and I’ve been working in it my whole career.”

That’s just not tenable for the new world, particularly with all the tools—agentic AI, generative AI, all of these things here now. It’s a much better operating mode when you can address the data gulch and the higher order of making better decisions as an organization.

I’ve continued to write about that. I have an upcoming book on it which will be published near the end of the summer. I’m very excited to talk more about the decision abyss and the underlying data gulch. That will be out in September.

James: And what’s the title of the book?

Keith: It’s called Conquering the Decision Abyss.

James: Oh, okay. There you go. Very excited about that—awesome, that’s fantastic.

I’m going to pivot again now and talk about some general industry trends within sourcing and supply chain. Have you noticed a lot of change in the past, say, five to ten years surrounding world events—things like regional conflicts, COVID—and how have those things affected the supply chain?

Keith: I would even go back further. I’ve been in supply chain my whole career. You used to have to explain to people what supply chain meant. I had to explain to my mother what a supply chain was, and nobody really cared. That’s okay—people go to the store, get a product. That’s the way the world should work.

For decades, ERP—enterprise resource planning—projects were prioritized. You did ERP consolidation projects, ERP rationalization projects. You did upgrades when it was on-prem, then you moved to cloud, and you had federated and non-federated ERP systems.

An ERP system is just a fancy term for a general ledger—an accounting ledger. It takes care of your finances, and that’s important. But organizations tended to invest in ERP systems and use bubble gum and duct tape to hold the supply chain together.

What COVID did: COVID is the great accelerator of supply chain innovation, because people finally realized how vital supply chain is to getting products to market and having robust, resilient processes.

If you visited a car dealer during COVID, there’d be one car on the lot. They couldn’t get microprocessors. There were all sorts of well-documented supply chain issues. COVID has been the greatest commercial ever for why you need a resilient, strong supply chain.

Those of us in the business have known this for a long time. We’ve been a bit of the canary in the coal mine, saying, “Your supply chain’s weak,” and you’d be surprised at some of the leading global brands that have inadequate and frail areas of their supply chain.

Now COVID has changed a lot of that. When I sit with CFOs or CEOs, supply chain is number one or two on their mind. We’re such an interconnected global economy. There’s a war in Israel. We’ve got unrest in other areas. We’ve got a U.S. president with tariffs coming up here and there. You see the frailty of the system in general.

It’s driven a lot of excitement for someone like me, because now you’re seeing real investment, real advancement, and supply chain and procurement being top of mind for C-suite leaders.

James: Would you say there’s been a shift from “just in time” to “just in case” with the supply chain?

Keith: It has been, yes. “Just in time,” “just in case”—great terms as the industry has evolved. But I’d go one layer deeper.

Companies have always stressed, in supply chain, the ability to forecast demand as a premium over quantifying the supply that’s needed. What COVID taught us—in some of the high-visibility examples—is that if you’re even one part or one plastic piece or one gasket short, you have work-in-progress on your balance sheet. It’s an unfinished good. That is crippling when you have a lot of WIP on your balance sheet.

If you’re producing something that requires 5 parts, 5,000 parts, or 50,000 parts, you can’t miss on one part. Think about that for a second. There are companies that make products and buy 200,000 separate MPNs in a given year. You can’t miss on the supply side.

For me, what I’ve seen—“just in time,” “just in case,” “real time,” all those are nice terms—is that companies that make products have always prioritized the demand signal more than the supply signal. Thank goodness it has now leveled out, because managing the supply side of the equation is equally as valuable as managing the demand side.

If you make phones, you’re trying to predict how many you’ll sell in a given market. That’s a worthy exercise. Equally worthy is making sure you have enough supply of all the parts to fulfill whatever the demand range is.

When you have 5,000 or 50,000 parts and one lead time is 24 weeks and one lead time is two weeks, that’s a problem. If it takes 24 weeks to get a part, your demand signal had better be really accurate. For people who don’t follow that, it’s like trying to throw an arrow at a moving dartboard.

On the demand side, you’re trying to slow the dartboard down. On the supply side, you’re trying to get enough arrows so you can throw a couple of times and hit it. For me, the important readjustment is that supply has become equally as important as demand.

James: Makes total sense, but it’s something people don’t think about.

So the last question I have for you before we wrap things up, coming back to AI one more time: how fast have you seen the rate of change in supply chain innovation grow since AI became readily available?

Keith: I’ll give you two answers. Obviously everyone’s talking about it. It’s huge.

I would say most of the AI I’ve seen in organizations has been in horizontal use cases—chatbots, things like that. There have been very few deeply vertical, specific AI use cases. Those of us in procurement and sourcing are a bit laggard in that regard. Some other areas of organizations are a little less mission-critical, and you can play around with AI use cases.

I’ve seen adoption in sourcing and procurement lag, and that’s not a bad thing. Sometimes you want to work out the kinks first. Every organization we deal with is talking about AI, investigating it, looking at it, evaluating it—with the right level of skepticism and awe at the same time.

AI is going to radically transform a lot. But if you look at technology innovation, things like this have come along before. There’s a wave of euphoria, then sobriety kicks in, and then the real building begins. I think we’re just getting over the euphoria phase and now the real work begins: changing human behavior with AI.

The technology is there—we’re already doing so much with it. The technology is not the problem. It’s the user pattern with the technology. It’s the behavioral aspects.

I can’t stress that enough. It’s not “Is AI good or bad?” or “Should we use AI, yes or no?” or “How much has it changed?” It’s really: how do you get a group of people who have done a job a certain way for so long to think differently? A lot of changes have to happen for that.

We’re on our way, but it will not happen overnight. Even though the tech is there, it’s not a technology problem. It’s a culture problem.

James: Makes sense. Well, honestly, it’s been a fascinating discussion. We could have kept talking for another hour easily, but we have come up to time. So I just wanted to say one more thing: if people wanted to keep up with LevaData—check out the products, the thought leadership—what’s the best place to do that?

Keith: Best place is to head over to the website: levadata.com. There’s a whole host of proof-of-concept options, ways to engage with us, ways to get free trials, ways to look at our software and engage with us. It will be well worth your time. Thank you, James. This has been great.