Stephany Lapierre ADT podcast cover
Episode: 130

Stephany Lapierre - How AI is turning uncertainty into opportunity in the supply chain

Posted on: 11 Apr 2024
Stephany Lapierre ADT podcast cover

Stephany Lapierre is the founder and CEO of the leading supplier data foundation platform TealBook.

In this episode, we discuss how the innovation in artificial intelligence can be leveraged to turn uncertainty into opportunity in the supply chain sector. We talk about the opportunities AI brings to procurement & the supply chain, as well as the challenges and key considerations for effective implementation. One of the key points that we highlight is the importance of responsible data management.

 

Links & mentions:

Transcript

"Don't get too hyped on technology. Just embrace it for being a tool to solve business problems and try to leverage what's already done because companies have raised and spend millions of dollars trying to solve these problems for multiple other organizations. And so, don't do this yourself, if you're not a tech company."

Intro:
Welcome to the Agile Digital Transformation podcast, where we explore different aspects of digital transformation and digital experience with your host, Tim Butara, content and community manager at Agiledrop.

Tim Butara: Hello everyone. Thanks for tuning in. I'm joined today by Stephany Lapierre, founder and CEO of the leading supplier data foundation, Tealbook. Today we'll be talking about how artificial intelligence can be leveraged in order to transform uncertainty into opportunity in order to drive innovation in the supply chain sector.

Stephany, welcome to our podcast. It's really, really great having you here with us today. Anything you'd like to to add before we dive into the conversation?

Stephany Lapierre: No, excited to be here. Thanks for having me. 

Tim Butara: Awesome. Then let's start with this. What would you say, or what have you seen to be the most typical use cases of AI in the supply chain?

And maybe, you know, maybe, are there some that are more surprising or less obvious ones that would be really interesting to our listeners? 

Stephany Lapierre: Well, I think, you know, AI is not new in the procurement supply chain world. I'd say it's not the most advanced in terms of having had the opportunity to leverage technology in the way that other functions, or even on our day to day we've been so fortunate to leverage.

And so I'd say it's, you know, maybe fairly, it's about a decade or two behind, you know, everything else. And so, but the exciting part is, even since COVID, the amount of sheer investment and attention that's been paid to supply chain procurement. And I think it was very apparent how dependent we are on having sustainable, scalable, resilient supply chains in our day to day live into business survival.

And so suddenly you have this massive investment. Going into our space, lots and lots and lots of technologies that didn't exist before addressing multiple different challenges within, you know, the procurement supply chain world. And you're in a space that's increasingly more disrupted and where requirements are changing, regulations are being added up.

You know, there's more demand for transparency. There's more demand for mitigating risks. There are more demand on commitment to sustainability and diversity, and also, you know, margins. And so it's a perfect kind of storm of opportunities to leverage technologies like AI. I'd say in terms of use cases, it's still fairly immature.

I think there's a lot of software companies who have been using aI for some time or machine learning within a software. And those technologies now are just advancing faster because there's more tools to be able to accelerate the investment and helping those functions to automate a lot of the workflows and the more mundane tactical tasks that they've been used to doing.

And then there's, you know, there's a lot of excitement obviously around large language models and ChatGPT. And so we've seen some really interesting use cases like drafting an RFP, right. Or an agreement or even writing content, you know, there's a lot being tested and I think it's going to become more and more sort of mainstream, but there's also a lot of concerns around, you know, sharing data are putting the organization at risk and sort of some around security. 

And there's also some concerns around having too many tools that have different types of AI and large language models across multiple functions, because, you know, in an enterprise one exposed to risk, but also, you know, you've got a lot of people using different tools, training, different models for different reasons. And so there's not a consolidated sort of intelligence that start to build for the enterprise. So kind of like a lot of, I guess my point is there's a lot happening. It's very exciting. 

There's more usage of AI and really good use cases and others that are just really, you know, are going to take time to mature because people are scared or unsure or organizations are putting pretty strict policies around how employees can use some of those technologies. And so, you know, I think it presents an amazing amount of opportunities for both sides. 

Tim Butara: Yeah, definitely. That was a really great way to kick off the episode. On the one hand, we have AI kind of not being a new thing. I think that that's kind of a theme of these episodes of ours revolving around AI, because everybody is kind of, has jumped on the hype train in the past year and a half or so, and we kind of forget that it's, you know, it's not a new technology, something that has been heavily utilized for the past 10 years, or in some cases, even more, as you said. 

And then on the other hand, we have these that you just mentioned, right? There's a lot of interest and a lot of excitement, but the best practices as to how to implement everything in the supply chain and kind of proper steps for innovation for AI in the supply chain are still getting figured out. And this is what we're going to likely see in the next few years, probably. 

Stephany Lapierre: Yeah, exactly. And I think the interesting part we talk a lot about is when procurement teams or supply chain teams are thinking about AI, might, and I don't know if we're going to go, that's one of the questions is sort of how, what the recommendation, but I always encourage customers or people within the industry just to take a moment and a day, even if it's a couple of hours or workshop on what task is being done, that's generating, you know, a pretty low value for the organization, but the work, you know, that's derived from, or the outcome of the work that's being done is really high impact.

And so, no, there's sort of a chart where you can sort of low value, low resources, but then you have some that are really resource intense and they're very manual and they're very repetitive and tactical and the value of getting the output is great for the organization. 

And so those are really great opportunities to start leveraging technologies like AI to see how can you actually propel some of those initiatives to get the better outcome without having to use so many, you know, people or resources or time to get the outcome, the output that you need. And so it's always, I find it a good exercise because there's a lot of those opportunities and you can start prioritizing how you want to automate some of those.

Tim Butara: So now maybe if we look at the other side, we talked about opportunities just now, what about maybe the challenges to effectively implementing AI into the supply chain? You mentioned before that one of the main challenges or one of the main risks is the concerns, the privacy concerns that people have, and what are the other major ones that would be relevant to people listening right now?

Stephany Lapierre: Yeah. I mean, that's my biased view is that in order to take the information that you're getting, especially if you're using something like the ChatGPT tools, that the data can be flawed.

And so if you don't have great data, if the data is outdated or it's, you know, sort of worldwide data that doesn't necessarily have, you know, a trusted source to it, then, you know, you are at risk of using the information in the wrong way or using the wrong information. And that's really dangerous in some cases. And it's also very bad for credibility in other cases. And so it still will require quite a bit of work. 

So the data foundation that you built to enable those tools to be more effective and trusted, I think is a really important thing to consider. And then I think that the risk, I find it's because of the maturity, there's an expectation, especially if you're coming from the business and you think technology is magic, which it is, but it's also, it needs time to mature.

And so I think sometimes there's an expectation that, oh, you know, it's not giving me the right answer, or it's not quite working the way it should be. And then suddenly people will consider this as a failed pilot or, you know, it's not, the technology is not working when in fact it takes time to mature, especially when, you know, AI is using a lot of data, it needs to learn and needs to mature in terms of developing, you know, models that are giving, you know, reducing the error rate.

And then you need to, you know, be... have some kind of patience and be willing to work aside, you know, some of those tools to refine, and then over time it will get better and it will bring more automation and more efficiencies and better output. 

So I think there's a bit of a lack sometimes of patience or even education on what does it require, you know, from an expectation perspective to invest in and stick with it versus sort of discounting it. And I'm talking just generally and you know, any tool - and I'm guilty of that myself, you know, I'll use, I use a large language model for fine yesterday to show my team and it gave me two wrong answers.

And I was like, oh, the data's not good. So I'm very guilty of this myself. Cause we're testing different things. Yeah. And so, yeah, I think it just requires patience and people on the business side that don't understand it have higher expectations often than the people that understand it because they know, you know, solving small problems in our case, like even, you know, normalizing a vendor master with all the suppliers and matching it to the right entity with the right name.

Seems really trivial, you know, but it's actually a really hard problem to solve. And so, if one of our customers have been trying to solve this problem for a long time, they are blown away by what we can do as someone that doesn't really understand would expect us to get this at a hundred percent. So, yeah, I think it's just education and understanding sort of what to expect, is really important.

Tim Butara: Yeah, I really like all of these points. So the responsible use of data is a prerequisite for responsible AI implementations. And people need to, you know, need to get on the same page about what responsible data usage means. And for that, you need proper education and patience with this education, because no education happens overnight.

And another point that I really loved in this answer, Stephanie, is that technology is like magic. And it like, even in the cases where that's true, you still need, like, you know, I'm sure that both us talking right now, as well as anybody listening right now has watched or read at least one piece of like fantasy fiction, like that involves magic or something.

And in every single instance, you need some kind of skills. You need to learn how to use the magic effectively. And it's the same way with technology, even if it's like magical or auto magical, I really like the phrase auto magic. 

Stephany Lapierre: Yeah, you got to control the power, but there's, two examples that I like, because I had to learn myself, I don't come from a technology background. I come from the biz. I was trying to solve a business problem and then discovered that you could use big data and AI to solve the problem, which was like, my brain exploded. 

There's a couple of situations and use cases that we've delivered over the years. One, it was, initially when we started building our dataset, we didn't understand if the data that we were creating on suppliers was good enough or not, like we didn't have a way, we were pulling a lot of data and then we built a search engine because we had enough context to build a knowledge graph to help customers find alternative suppliers. 

And so what we found is that at the, especially at the beginning, you know, sourcing manager would go and click to find a supplier and it would not give the right recommendations and then they would get, you know, or they would see something in the profile, I know Mary's been gone for five years. And so your data is bad. Right. 

And so, but it's the customers who work with it, who gave us also feedback because we need to refine the data. We need to refine how they're researching and the type of output they're looking for, and then the quality of the output and our customers, I always remember.

You know, you had a client on the same day that was like, you know, it gives me bad recommendation. I think your data is not ready. And then you had another client who had a different mindset. And she said, listen, Steph, like, the work that I've done took me four hours, so it was not perfect. I had to toggle with the search words.

I had to filter, I had to sort of qualify, but it took me four hours to get back to my stakeholders. Normally it would take me two weeks. Right. And so the win was that for the fact that I was able to get back to my team in the same day, gave me so much credibility and enabled them to move faster versus waiting two weeks.

And so, yes, it wasn't perfect on the first search. She had to work with it, but because she worked with it and then over time, it just got better and faster and faster to a point where she was comfortable actually giving the tool to her stakeholders so that they could search for information themselves, then it's true enablement, but right.

And then someone completely discounted. So start at zero again, probably went to look for other solution, you know, versus sort of sticking with it. And another example, if I may, is one of the use case that we start solving for is it's really hard to, it's easier to find, first of all, you'll know the suppliers you work well and are very strategic and you spend a lot of money with, so those are, you know, yes, you still need to collect information, maintain it, but the value of AI is looking at lots and lots and lots of data.

And when you have, you know, tens to hundreds of thousands of small suppliers that you don't spend a lot of money with potentially, they're privately owned. So the data is not as easily to access. It's not just publicly available in the way that publicly traded companies are, but that's where you have the most change.

That's where you have the least likely suppliers will engage with a portal. So it's hard to maintain, collect and maintain information. It also introduce the most risk. And it's also where you get the most opportunities for innovation. Cause there's a lot of startups. And then if you add all of the spend, all of this data, it's huge levers for savings for hitting ESG targets and things like that.

Anyway, so our clients were really pushing us to get better at understanding small privately owned businesses and specifically ones that were certified small and diverse. And so that was a use case that we solved about four or five years ago. But I always remember the first exercise. The client said, can you help us automate the reporting and help us find net new diverse businesses that are certified and our own vendor master?

And if you could prove this out, it would be a huge win. And so we asked, like, what would, like, can you give us, what would be a win for automating reporting? They said, if you could automate by 25%, so which was totally reasonable. And the second one is like, if you could help us find 50 to a hundred net new diverse businesses in our own vendor master, it would be a huge win.

And we had no idea, like we didn't know. So we start looking for sources of data, start pulling it together. Two weeks later, we come back with our report. And their response was like, it's no better from a total spend. You didn't find more spend. And, you know, it automated, it was close to 25%, but the big thing for them is that we had not found more spend than their incumbent provider, which is a software company that had been around for 20 years.

And so their reaction was like, it's not working. Our reaction was like, hold on, are we the same? And they're like, yeah, it's pretty much the same. Like, so you're telling us that we're the same within two weeks of starting to work on this, then the company has been around for 20 years. And then they're like, oh yeah, you're right.

Like, give us another two weeks. We're going to go back. And we ended up, you know, automating 75% of the reporting. We found 800 net new businesses in their own vendor master that were certified that they hadn't caught. And so. You know, it's so much about storytelling and expectation and setting the right expectation.

And because you could have completely two different reactions and output, but it's always again, putting in context that this technology continues to mature and evolve. And if you make any incremental improvement to what you're doing today, and you're able to track that progress over time, that's where the patient has to come in and work through it so that eventually it becomes incredibly efficient and does things that, you know, there's no human that could do it as at the speed and scale that, you know, automation and technology could do it like AI.

Tim Butara: I really love this example and it showcases kind of the bias of experience, right? Because we know that something has been working, we don't even consider how like a new approach then relatively compares to that as opposed to in absolute terms, right? They probably, the company, their initial reaction was based on like the absolute improvement, which was, you know, 0% because it was the same, but if you looked at relative improvement, what you achieved in two weeks versus what the company achieved throughout its entire lifespan, then that's a much different data point, basically.

Stephany Lapierre: Yeah, I think it's just... it's storytelling. It's setting the right expectation and reminding... often you have to remind customers of what we set out to achieve at the very beginning, what would be success? And let's make it something that we can actually high five and two months or three months from now.

And that's... remember, because as soon as you know, you hit certain things and suddenly ideas are flowing, expectations grow, you know, we forget what we were set out to do. And so I think it's so important to kind of set the right expectation and realistic goals at the beginning, communicate those goals really effectively.

And then once you hit and you show progress towards that goal, then set a different, another goal, right? Another stretch goal, to be able to continue to work towards a bigger, you know, if you want full enablement, you don't get full enablement, you know, on day one, it takes time. And so I, I think that's probably the biggest.

It's gap or challenge within the business sense of procurement and supply chain is setting the right expectation and then see that sort of incremental improvement achieve over time and have the patience to work through it. And it requires the right mindset and certainly you have the right change management through skills and yeah, mindset from a team to be able to be good partner through that process.

Tim Butara: Well, I'm sure that people listening right now will be much better equipped to do so, to do this properly in the future. Loving this conversation, Stephany, loving the advice. And I'm also interested, since you're, you're hands on with all this, you work with clients. What are some of the biggest trends in terms of AI and the supply chain that we're likely to see this year? So in 2024. 

Stephany Lapierre: Well, so I'm going to be super biased about this, but I do think data, data, good quality data is foundational to everything we do. Everything we do leads to a supplier. So if you don't have the information about your suppliers, enriched, accessible, actionable in a way that can enable your systems and your processes and your workflow and the people making decision, it's going to be a hard hard journey, because disruption is not going to slow down.

There's not going to be less regulations. There's not going to be less demand for automation and most, you know, we're looking at in the market, multiples of companies that are operating more efficiently or significantly higher than ones that are not. And so there is a responsibility within procurement and supply chain.

Not only to help the organization achieve its goals and build sustainable, scalable growth, but you also have to build that more efficiently. So you have to have this mindset. And so I would say my bias is towards data, like AI can, AI is amazing if it has good data, but AI can also be applied to getting better quality data.

That's how we're using it. You know, it's, it's gives us an opportunity to give a foundation to companies that they can build. And instead of relying on people and suppliers and putting information multiple places and try to make sense and consolidate and normalize and, and, you know, batch this data, they can actually build a foundation that provides consistent high quality or high trusted data across all their systems and tools.

It's pretty game changing. And then that allows you to leverage, you know, the tools that have AI. And what we're seeing as a trend is a lot of companies are now the ones that are working closely with the CIOs and the IT team. They're looking to leverage enterprise data lakes, data warehousing that they've already invested, and rather than having multiple tools with different analytics, they're looking to consolidate analytics into a Power BI or a Tableau and then leverage the large language models on top of that.

And so that's a huge opportunity for procurement and supply chain to master the supplier data of ownership and make it more actionable because not only procurement and supply chain will benefit from it, but it will also enable all the other functions that also need this data, right. To get access to the same information and, and bring a lot, make the organization a lot more intelligent in terms of, you know, capitalizing on the data.

On their buying power and leveraging this massive asset, which is your supplier base to achieve your business goals. And so, you know, you can enable finance compliance, third party risk, ESG team, even sales teams, if they, they can use your suppliers as pipeline or ways to better negotiate with customers.

And so endless opportunities, you know, again, I'm super biased and using AI for data, but that's, you know, and then data, data can enable all these other fantastic things that we can achieve. 

Tim Butara: I love the part about consolidation of data and how important that is because basically, you know, even with a small amount or a small degree of complexity, unified data is key to success in basically anything.

But then as you add disruptions, as you add people, as you add departments, as you add partners, as you add vendors, and as all of this is then also more and more impacted by regulations, by disruptions that are often unpredictable, unified data. I mean, data that's not unified in that case, just, it's just a recipe for disaster, basically. 

Stephany Lapierre: Frankenstein architecture, swivel chair between systems. It's people making, you know, spending way too much time doing the same things that doesn't really deliver value. And ultimately you're not helping your enterprise, your organization get a lot more intelligent. And so I think it's a 2024 is a huge opportunity for procurement and supply chain to really take, master this, you know, and position itself to be the enabler of better information on the buy side. And so, yeah, huge opportunity. 

Tim Butara: Well, it'll definitely be exciting to see how it plays out. And Stephany, I really enjoyed this conversation. Just before we start wrapping it up, I have one final question. So far, we mostly focused on AI and on data, which is also closely tied to AI. But I'm wondering if there are any other important factors or new technologies that are driving innovation in the supply chain world.

Stephany Lapierre: I'm sure they are, you know, we leverage all kinds of technology. AI is just one of, and we're not building AI. We're not building models anymore. We used to, and now we just leverage. There's so many amazing tools that we can connect together. And I think it's more of that mindset is having teams leveraging what exists and be able to put it together in a way that, you know, make the organization or your projects much more efficient.

Yeah. I mean, trends come and go, right. They're all at the end of the day, they're just tools. And so, and you need to stay on top of the tools that are helping you do the job and and, and make it better and more efficient over time. So I can't think of, you know, I'm so such embedded in this world right now.

Like. I heard yesterday, someone talk about blockchain. I hadn't heard so long, but I remember like two years ago, three years ago, it was a massive hype. And so don't get too hyped on technology, just embrace it for being a tool to solve business problems and try to leverage what's already done because companies have raised and spend millions of dollars trying to solve these problems for multiple other organizations. And so don't do this yourself, if you're not a tech company, if you're a bank or you're in retail, you know, focus on your business and leverage the technology that exists to deliver a better outcome for your customers and your team. 

Tim Butara: That was a perfect tip to finish off this awesome conversation on Stephany. Just before we jump off the call, if listeners would like to reach out to you or connect with you or learn more about you or Tealbook, where would you send them to? 

Stephany Lapierre: Yeah. So of course you can come to our website at tealbook.com. We have a resource library with lots of presentations and webinar articles on the topic.

And you can follow me on on LinkedIn. It's Stephany with a Y and yeah, I do post a lot. So if you're, if you're willing to get my posts, if you're interested in this topic, yeah, please connect. I'm always, I'm always open to having a good conversation. 

Tim Butara: Awesome. Well, thanks again, Stephany. This was fantastic.

Stephany Lapierre: Thank you for having me. 

Tim Butara: And well, to our listeners, that's all for this episode. Have a great day, everyone. And stay safe. 

Outro:
Thanks for tuning in. If you'd like to check out our other episodes, you can find all of them at agiledrop.com/podcast, as well as on all the most popular podcasting platforms. Make sure to subscribe so you don't miss any new episodes. And don't forget to share the podcast with your friends and colleagues.

Listen offline

Share