Razat Gaurav
Episode
6

Innovation in supply chain decision making and design has flourished in recent years. The availability of high-quality data coupled with an urgent need to respond to disruption has fostered rapid change in how organizations coordinate, cooperate, and improve their operations. Join LLamasoft CEO, Razat Gaurav in conversation with MIT CTL's Matthias Winkenbach to get a glimpse of how the world's most successful companies are organizing to meet their decision and design challenges through data democratization, the use of micro-apps and by engaging in a federated solutions approach. Learn more at MIT CTL. Visit the MIT Computational and Visual Education (CAVE) Lab.

Transcript

Host:
Welcome to MIT Supply Chain Frontiers, where we discover the future of global supply chain education, research, and innovation. Today, CTL's Matthias Winkenbach speaks with Llamasoft CEO Razat Gaurav about supply chain, decision making and design.

Matthias Winkenbach:
Thanks for joining us today. Thanks for having me Matthias. Today's topic is supply chain decision making, supply chain and design, but why don't you start by giving us a quick introduction to the company, but also to your background and what got you excited about this industry?

Razat Gaurav:
Absolutely. Uh, you know, LLamasoft we're a supply chain analytic software company have been in the supply chain design space for a long time. And then in the last three, four years, we've really been expanding into a broader set of supply chain decision-making capabilities. But what excited me about Llamasoft with three things, one was the underlying data model and the ability for you to capture the overall end to end supply chain into one single data model. Uh, the second thing I found was just the algorithmic horsepower, you know, and the third aspect that I found fascinating was the company had been investing in a low-code no-code app development platform, uh, through a micro app based architecture.

Matthias Winkenbach:
Fascinating. Um, before I get into the technical details in these days, we can't really talk about, uh, supply chain decision-making without first touching upon kind of the elephant in the room, which is the implications of COVID-19. What changes are you seeing in the way, but also the magnitude at which people are thinking about supply chain design as a means to overcome the crisis. And, um, how has their attitude towards supply chain design changed, uh, compared to the time before COVID?

Razat Gaurav:
Yeah, no, that's so that's a good question. What we found was that every weak link in the supply chain, uh, just got put on a magnifying glass with the COVID 19 disruption. Uh, obviously, uh, the first few months were very much focused on the firefighting mode, uh, in terms of responding to different types of disruptions. I think from there as companies sort of try to stabilize their, their, their operations and they, everyone started sort of taking a step back and thinking about what are some of the structural changes we need to make in our supply chain. And that's where fundamentally rethinking the design of the supply chain is something that, uh, many, many companies are beginning to reevaluate. And, uh, frankly, many heads of supply chain are being asked to do that by their CEO or by their boardrooms.

Matthias Winkenbach:
What would you say distinguishes, let's say the best in class companies out there from maybe those that aren't as prepared to actually see supply chain design is such a valuable tool.

Razat Gaurav:
Yeah, that's an important point. You know, I think the first piece is, you know, those companies that have invested in the data foundation in being able to, uh, get all the relevant data from across all the various systems in your, in your landscape, because, you know, supply chain data doesn't reside in any one system supply chain data resides in tens of hundreds of different systems in most, uh, large manufacturing and retail organizations. And so being able to cure, rate that data and, uh, have a system of reference that you can keep refreshed on an ongoing basis is very foundational to be able to do supply chain design and to make decisions that are data-driven and fact driven, and to also be able to leverage the power of algorithms you can't use advanced and sophisticated algorithms without having the underlying data. So I think those companies that hadn't invested in that foundation and that infrastructure were able to very quickly, uh, use the, the, the, what if analysis and the scenario planning capabilities to understand the impacts with the COVID-19 disruptions and then to make the right decisions to take the corrective actions. Of course, some companies had a fundamentally, uh, more agile supply chain design, where they had more optionality, right? So if, if you inherently had more optionality in terms of where you could source from when China shut down, you could easily, uh, you know, uh, tap into other sources of supply, like could be Vietnam or Philippines or India, right? Uh, some companies did not have that. Many companies still had a lot of single points of failure.

Matthias Winkenbach:
Do you think the companies that have those particularly agile and kind of responsive supply chains, what differentiated those, that they just have better tools that they have better talent? What do you think was kind of the key driver of their success in this area?

Razat Gaurav:
I think, you know, some companies have taken the whole area of supply chain risk very seriously, and some companies haven't and those companies who haven't taken this area of supply chain risk, uh, as a core part of making design decisions, what caught on the wrong footing, we found that companies, their backup production plant for a facility, let's say in Rouhani, maybe another production plant in a different part of China. When we think of supply chain risk. And we think of how can companies get more optionality? This has to be thought in a much more profound way where you need to think of optionality across different regions, right? The other important thing is thank God we have something called buffer inventory, the supply chain, those industries or companies that were extremely lean at the same time did not have much in terms of optionality for sources of supply. They were probably affected the worst, right? So the balance between cost service and risk is something that many companies are going to grapple with in very profound ways going forward, because COVID-19 has shown that, uh, just, just trying to minimize inventory and just trying to minimize working capital, lots, trying to minimize cost sometimes increases your risk factors dramatically.

Matthias Winkenbach:
Would you argue that companies where supply chain design is primarily driven by a limited number of supply chain experts rather than being a board level topic? Do you think these companies are more prone to running into these risks because they do not have that view?

Razat Gaurav:
I totally agree with that premise because in our estimation for over 60, 70% of manufacturing and retail companies, uh, supply chain design is a very asymptotic process. They reevaluate the design maybe once in two, three years. And the problem is the business is changing a lot more rapidly. It's like concrete. You, you make certain decisions on your flows, on your locations, on your, you know, we like to think of it in terms of nodes flows and policies. And these three elements, the nodes being the locations flows, meaning product flows, uh, policies are policies. There's many, many different policies that govern the supply chain like, like inventory policies, replenishment policies, mode policies, and they typically get established during an initiative, or when you're trying to enter a new market or you're trying to launch a new product. However, they set in like concrete, okay. And they become extremely rigid and no one is going back and reevaluating and requesting the fundamental assumptions that were made when those decisions were made for these nodes flows and policies, those companies that were not doing this on an ongoing basis where the supply chain design competency was not something that was systemic.

Razat Gaurav:
It was not something that their executive team was focused on what caught on the wrong footing, because, you know, uh, when you have a very rigid system, it's very automated, brutal system. You have not factored in, uh, various risk factors that you typically would unravel if you, if you were doing supply chain design more on a, on an ongoing basis, but in order to do that on an ongoing basis, first as you need the data foundation, second, do you need yes, some tools, but as importantly, need talent, you need resources who are capable of understanding your physical operations in the supply chain, reflect them within the constructs of a digital data model. And then to be able to analyze different scenarios, using a variety of advanced algorithms that skillset of a, of a supply chain designer or a supply chain modular is a key skill set that needs to become a lot more systemic in many organizations. And, uh, and of course that's also an area where our partnership with MIT is, is, is really important.

Matthias Winkenbach:
We've seen an hour and also the need by various different industries actually to invest, not just in domain specialists or methodological specialists like pure data scientists, for instance, but we've seen an increasing need for what we would usually call the data translators, which actually brings me to another point that I wanted to touch upon. Namely, the paradigm shift that we're seeing in supply chain decision-making, which is by the way, also the core focus of the newly created cave lab here at MIT. So from our point of view of supply chain design is becoming more and more data and analytics driven. And at the same time, it's becoming more of an ongoing activity across various stakeholders within the organization, rather than just a periodic one-off effort by a relatively small group, from your point of view, to support this shift in the corporate perception of supply chain design, which tools and what kind of talent should companies invest in,

Razat Gaurav:
We found is the need for doing that type of analysis, where you are trying to not just look at the current state of your operations and your network, where you are trying to understand what is the impact of changing various degrees of freedom is something that you need, not just for making big strategic long-term capital intensive decisions. You need that same skill set to even make more near-term decisions, you know, like decisions that could be for the next quarter or the next month, sometimes even for the next week. And this type of knowledge and insight that is data-driven is what is required. Not just for people in the supply chain organization is also required for people outside of the supply chain. The ability to leverage depth of capabilities in a simplistic way is something that we have been really very focused on. We think of that in the context of like micro apps.

Razat Gaurav:
So think about your mobile phone. All of these apps have very, very sophisticated engines running behind them. We had to have a lot of depth of technology running behind them, but me as a user, I don't need to go through, you know, uh, uh, five days of training and have all kinds of understanding of all those technical details. I just look at leveraging micro app for the purpose that I'm looking for, a specific type of analysis or insight or information can be brought, I think, to supply chain design, and decision-making through a micro app architecture, which is what will enable the democratization of these capabilities and have a much, much bigger impact than only two, three, four, five people in an organization being able to do that type of work.

Matthias Winkenbach:
That's very interesting. So one thing that I wonder in that context is like, how do you boil down the inherent complexity of some of these decisions, such that they are actually digestible by a non-technical user? What do you see as the main lever is to actually simplify this down to a level where an average user can actually turn these insights into value without oversimplifying

Razat Gaurav:
No problem. You will still need the really, I think of them as a fighter pilots, you know, who are the deep, uh, you know, modelers and data scientists who can model the complexities of the supply chain, right? And to reflect that with all the different interrelationships, you will still need that, but then they can serve up the application in a very narrow and purpose-built way to the Mo to the broader users, right? So the broader users are just going to use it for what they need it for. So for example, if I'm, uh, responsible for, uh, my transportation fleet routes and every day, I need to make sure that I understand what my shipment plan is going to look like for the coming week or for the coming few days. I don't need to know all the details of how the model has been set up, which is leveraging very sophisticated multipack, multi drop optimization, and looking for all kinds of constraint representation that can be done by a small group of, let's say, a center of excellence kind of resource base.

Razat Gaurav:
But then that app is served up to all the different people who are in the warehouses, who are at the dock doors, who are, who don't have all those detailed data science skill sets in an app where they can understand with one push of the button, what the shipping plan is for that particular day or for that particular week. Right. So we had to think of this in a federated concept, reflecting all the nuances of the network that requires depth of skills and specialized capabilities. Once I have set up that baseline model, can I create an app that allows you to run different scenarios for let's say inventory or a changing demand picture, or, uh, you know, just changing like the operating hours of facilities or adding a new note to the supply chain? Can I simplify that for the broader organization? Absolutely. Yes, but it has to be done in more bite-sized and more narrow scope solutions, number one, and then number two, it still needs to get supported in the background by the heavy duty modelers who will be able to understand all the different complexities. So it's the federated architecture that is very key, both organizationally and from a technology perspective.

Matthias Winkenbach:
So there's that I guess you would agree that for most companies out there it's becoming less and less difficult to obtain large amounts of high quality data, both on their own operations, but also on the environment that they're operating in. And on top of that, there are people like us who keep inventing ever more powerful algorithms, ever more sophisticated tools for supply chain design. So if you think 10 years down the line, do you think humans will still play a crucial role in supply chain design decisions at all? Or will most of this actually be fully automated?

Razat Gaurav:
Look, I think there's a lot of applicability for using AI techniques to automate certain decisions, right? But will that take away the need for human judgment and human involvement? Absolutely not. In our point of view, now, the role of the human persona is going to be different in the world of AI than what it has been traditionally, because you are able to, uh, leverage, leverage new forms of data, and you are able to get new forms of insights, and there will be a subset of decisions that you may be able to automate. But when it comes to a judgment that we have not yet seen AI techniques in the supply chain context, being able to replace a human judgment, that's one aspect. The second aspect is you still need humans to feature engineer. These AI algorithms. You'd still need humans to be able to understand what data sources to tap into.

Razat Gaurav:
You still need humans to be able to change the parameters as your business changes as your operating model changes as your business model changes. So the, the role of the human person may evolve and change in supply chain design, but we think absolutely there will be humans involved. And many of these decisions that you're trying to make, especially in supply chain design, when you have so many different degrees of freedom, they require discussion and debate in the broader organization. Sometimes they involve many, many different parts of, uh, you know, many different functions of the organization because they may require big capital investments, or they may require big changes to your overall network or to your, to the way that you're thinking about policies like inventory or fulfillment to customers. So they involve many different parts of the organization, even beyond supply chain. That dialogue, that discussion that collaboration we think is still going to be very central to how design decisions get made, as opposed to automating everything under the face of the sun.

Matthias Winkenbach:
And do you think those micro apps are probably a way of crowdsourcing that highly fragmented and highly decentralized knowledge about very specific aspects of the supply chain? So do you see this basically also as a means of improving the algorithm by basically crowdsourcing the human knowledge?

Razat Gaurav:
Absolutely. And that's where, you know, Google maps, for example, has some of the most sophisticated algorithms underpinning it, right? But for me as a user of Google maps, I don't need to know anything about those algorithms. And that's what allows us to scale the impact of Google maps to millions of people. The same thing applies when it comes to supply chain decisions and these micro apps, and being able to democratize the impact of sophisticated decision making capabilities, analytically oriented decision making capabilities, but putting it in the hands of many, many more people, you know, typical CPG organization, large multi-billion dollar CPG organization, doesn't only have a small supply chain organization. They typically will have, you know, tens of thousands of people in their supply chain organization. If they can have micro apps that allow them to function better, that give them better information, better insights allows them to run different types of analysis in a simplistic way so that the sophistication is masked for them. That's where I think we'll have to, we'll be really be able to move the needle. And to your point, that will also allow us to make the algorithms because a lot of these are constantly continuously learning algorithms become a lot smarter over time, uh, whether it's algorithms over the parameters or whether it's the features that we are engineering around these algorithms, all of those become smarter when there's broader usage of these capabilities across the organization.

Matthias Winkenbach:
So you mentioned earlier that, um, you see this idea of micro apps as basically being something that pretty much everyone within the organization could run, let's say, on their smartphone. One thing that we've been exploring in the cave lab at MIT, which has been built around rethinking the way we do supply chain decision-making in the future is interfaces. Like how do we actually convey this massive amount of data and this massive amount of complexity in many supply chain decision problems to a user, but also how do we kind of elicit from that user rich inputs? So what is your vision for the future interface to basically a unified supply chain decision making tool?

Razat Gaurav:
This is a fertile ground for, for research. Um, organizational decisions are inherently complicated and they very rarely just involve, uh, one individual or one function. Most especially supply chain design decisions. They typically involve multiple people, multiple functions, how those different people and how those different functions can converge on a common vision of the truth is something that I think is still a work in progress, right? And how you help companies visualize the data and how you enable these practitioners to be able to digest the essence from that data is very important. One thing we are investing a lot in, and this can also be an interesting area for us to collaborate with the cave lab is, um, how can we leverage AI to, to expose questions, to expose or suggest, uh, ideas that may not be very, uh, relevant at the surface. So based on a triangulation of various types of data, how can we focus them on the three things that are going to be most impactful for whatever area that they're focused on could be inventory.

Razat Gaurav:
It could be a demand, could be, uh, you know, locations could be transportation, could be capacity, production, capacity. How do we, because there's so much data to digest and, and there's always new forms of data that are coming in, especially now from edge devices, from IOT sources, like weather, like, uh, satellite images on ports, Google mobility, indices that are being published, this more data than we know what to do with. So part of the trick in synthesizing that data to help these organizations make smarter decisions is going to require for us to be able to say out of all the different things that data is exposing here are the three things, the four things that are most relevant to what you need. Now, there's a lot of sophistication that needs to happen in the background, and that needs to be a common dataset. I mean, it sounds like a pedantic thing, but today we find that there aren't even consistent data sets or cases consistent kind of assumptions on the supply chain that are made across strategic tactical and operational time horizons. That's why Microsoft Excel continues to be the most dominant supply chain decision making tool in the world, getting to a common form of data, uh, and a common foundation of that data, but then being able to use the new forms of analytics to synthesize and shortlist in a simplistic way, that key insights that are relevant to the types of decisions is, is going to be very, very key part of, of how, uh, this decision-making can become more effective.

Matthias Winkenbach:
So you already hinted towards it. I mean, Llamasoft just recently announced that it became both a strategic member of MIT CTL supply chain exchange program, but also, um, a strategic collaborator of the MIT cave lab from your point of view, like, what do you see as the short-term gains from this partnership that could also help, um, CTL its partners, but also your customers? Um, ideally both of them to learn something about how to actually mitigate the current crisis

Razat Gaurav:
In the short term. Uh, we, we really are looking to take advantage of the cave labs infrastructure, but also, uh, access to some of the, the, the team members at MIT CTL to host visioning sessions, uh, with a lot of our customers, you know, our customers are going through a timeframe where they have to navigate and make a lot of very complicated decisions that require complicated. Trade-offs the cave lab really provides a great setting for, for us to be able to facilitate these visioning sessions, using data, using insights, and bringing the teams together to make these complicated decisions in the most informed way and to really redefine and reimagine the way the decisions are getting made. Um, I think, uh, in the, in the term, I see, uh, a lot of opportunities for us to also do some joint research. Uh, there, there are many areas, uh, as we think, especially in the area of supply chain design and the area of supply chain risk in the area of, uh, how organizational decisions can become more, uh, more seamlessly, uh, structured, uh, using technology and using data is a topic that is very common to both, uh, MIT CTL and the cave lab, as well as, uh, soft.

Razat Gaurav:
So, uh, I'm hoping that this, uh, this partnership would evolve to also become a partnership where we can do some joint, uh, research and co-innovation as well.

Matthias Winkenbach:
That sounds great. And we're certainly looking forward to that. Unfortunately, we're running short on time. Um, so I have like one final question for you. Uh, imagine it's 2030 and the next pandemic, um, hits the globe, will the supply chain world be better prepared next time?

Razat Gaurav:
Look, I really hope that this pandemic as, as disruptive and as appalling and tragic as it has been, can also be leveraged to make some of the key changes and investments that are required to architect and design a more resilient supply chain. Many times companies have gone through disruptions. None of them have been of this magnitude or scale. And ultimately the impetus for can not just come from the supply chain organizations. The impetus for that has to come from shareholders and from boardrooms of companies and CEOs and CFOs, because this complicated trade off between cost service and risk is something that, uh, will require some tough decision-making and some significant investments in order for you to be better prepared for the pandemic. But I think there were many lessons learned all the way from the availability of critical medical supplies and PPE equipment to how, uh, a coordinated response should be structured, how a more coordinated testing infrastructure can be structured as in when vaccines become available, how should they get distributed?

Razat Gaurav:
So there's a lot that, that, uh, has been exposed that we need to work on still, right? And we need to have the right processes and the right governance and the right systems that can enable a more coordinated response. Also, I think in many companies and industries, we have exposed to certain single points of failure, right? As you map, you know, the various tiers of the supply chain, you know, you start looking at your tier two or tier three suppliers. A lot of, uh, concentration has been exposed in certain industries. And, and that's where I think the there's a, there's going to be a lot of appetite for creating optionality in the supply chain for creating a different level of approach to making, sourcing decisions, production decisions, routing decisions, and, and so have, have all companies already made those changes already. I don't think so.

Razat Gaurav:
Most companies haven't, I think most companies are still in that journey and they're trying to grapple with some complex and tough decisions to make, but I really hope that this crisis becomes also a, the, the, the big trigger point for companies, uh, taking supply chain design seriously and architecting and investing in the talent and the tools and the processes that make it an ongoing process. That's really key, uh, for us to be better prepared for the next pandemic. All right. So let's, let's see, what's going to happen, but, uh, thank you. This has been a really fascinating conversation. And, um, we're looking forward at CTL to, uh, keep working with you guys on the research sites to advance some of these topics that we just discussed. Thank you. That's great. Good talking to you, Matthias. Thank you.

Host:
All right, everyone. Thank you for listening. I hope you enjoyed this edition of MIT supply chain frontiers. My name is Arthur Grau communications officer for the center. I invite you to visit anytime at ctl.mit.edu or search for MIT supply chain frontiers on your favorite listening platform. Until next time.