Uptime: Dr. Selene Silvestri

'Minutes, Not Days': 

Dr. Selene Silvestri on AI Tools for Supply Chain Decision-Making

"Driving the Future of Global Supply Chain Integration" —that's the motto that greets visitors to the website of MIT's Center for Transportation and Logistics in Cambridge, MA. Through research, education, and stakeholder engagement, MIT CTL doesn't merely study supply chains—they advance what’s possible and turn insight into action.

Our guest for this episode is Dr. Selene Silvestri, a research scientist at the MIT CTL in the Intelligent Logistics Systems Lab. Silvestri's current research addresses supply chain network design and optimization, to help companies improve decision-making in their supply chain.

In Paper360°'s May/June issue, readers learn about one of the Lab's most exciting new projects: the GENESIS platform, developed in partnership with Mecalux, a warehouse technology and intralogistics software enterprise. Silvestri discusses GENESIS and much more in this engaging episode of UpTime.

"What GENESIS allows you to do is solve scenarios in minutes," she tells listeners. "And if you can solve scenarios in minutes, you do not only solve those that you want to validate. You solve scenarios that you want to explore, right?"

Stay tuned until the Lightning Round, when co-host Mark Rushton asks Dr. Silverstri about the biggest changes she has seen in the packaging industry over the last 10 years.

Learn more about MIT's Center for Transportation and Logistics and its many specialized labs—including the Computational Analytics, Visualization & Education (CAVE) Lab that Dr. Silvestri references in this conversation.

 Discover Workforce & Leadership Development opportunities that can help your mill or plant build, upskill, and advance your supply chain employees and teams.

 Access more resources through the TAPPI OnDemand platform:

• Paper360° Archives, including the complete May/June 2026 issue.
• UpTime episode featuring Wesco's Jefferson Smith and Matt Lasnicki on navigating supply chain challenges.
• Mark Rushton's article Saica in Spain: A Walk Into the Future of Mill Logistics.

Transcript

SPEAKER_02 0:03

The pulp, paper, tissue and packaging industries are always growing, changing, and surprising us every day with something new.

SPEAKER_03 0:11

And what drives that dynamic growth? Our industry's people. Hi, I'm Jan Bottelari, editor of Paper 360.

SPEAKER_02 0:20

And I'm Mark Rushton, editor of Tissue 360. Welcome to Uptime, a paper and packaging podcast. Each month we will connect our readers and listeners with some of the best stories and articles from Tappy publications.

SPEAKER_03 0:37

A dose of positivity to fuel your workday. That's Uptime. Let's get started. Hi, I'm Janne Badaleri, and I am here with uh Mark Rushton.

SPEAKER_02 0:51

And uh I'm actually coming to you from Helsinki at the moment where I've attended Pulp and Beyond, which has been very exciting.

SPEAKER_03 0:58

Oh, that's terrific, Mark.

SPEAKER_01 1:00

And our guest today is Selena Silvestri. Hi, everyone. Uh I am Selena, a resource scientist at the MIT Center for Transportation and Logistics, and I'm part of the Intelligent Logistics System Lab here at the center.

SPEAKER_03 1:13

That's terrific. Thank you so much for joining us, Dr. Silvestri. And you know, it's it's amazing when we first started working on the article that appears in our major issue of Paper 360. I was surprised and rather delighted to learn that a university like MIT has a lab devoted to logistics, right? Logistics systems, supply chain management. Can you tell us a little more about your work there and about the lab and what you all do?

SPEAKER_01 1:49

Yes, absolutely. Um, so maybe I'll start by telling you something about the research center itself, so the Center for Transportation and Logistics, um, to which I belong. Um, the center, as I was mentioning, is a research center, and our focus is um improving supply chain efficiency. Um, this, in simple words, means um we um try to understand how to design supply chains and how to manage operations in the best possible way, um, meaning making them faster, making them more uh cost-effective, reliable, resilient. Um, and this goes from network design decisions, where do I position my facilities, where do I locate distribution centers, how transportation decisions, so how do I flow goods um across the supply chain network, inventory management, uh, which is the right place and the right time to store my inventory across distributed warehouses, and last mile, uh last mile delivery, which is actually, I like to describe it as the moment in which supply chain uh meets the final customers. Uh, so we order something online and that something is delivered at our door. And that's actually the most expensive piece of a supply chain, the most complex and expensive. Um, what makes our lab, our center, I would say, very um special and unique is the fact that we do not conduct research for the sake of doing research and um developing innovative solutions. We do that in partnership with companies. So we want to solve real world supply chain challenges. Um and this is the goal, um, the mission of the entire CTL. But the way we operate is we are divided into um different labs, and different labs look at this scope um from a different angle, uh, which um it's usually related to the domain of the lab. So, for example, we also have a humanitarian logistics lab, um, a freight lab, a sustainable one lab. Um, the lab to which I belong, uh, as you were mentioning earlier, is called Intelligent Logistics System Lab. This lab was launched, uh we launched it like one year and a half ago, more or less. And the reason um behind this new lab um is that we recognize the need of um leveraging not only operational research methods, which is what we have usually usually historically used to solve challenges in supply chain, but also new technologies and the power of AI and machine learning to embed them, integrate them with operational research techniques, with the idea of tackling large-scale and complex and dynamic problems. Um, if I can add something on that, is um there is another lab which is called the Cave, and we collaborate a lot with the Cave Lab because we we might be able to develop the most sophisticated uh solutions in the sector, but if we do not make those solutions available to final users, to technical, but also to business users, these solutions don't really, I don't want to say are not needed, but their power is it's somehow lost, you know. Um, the cave lab is actually that mission. So it was launched several years ago with the idea of saying, okay, let's transform analytics solutions into tools that business users can actually use and can um benefit to take decisions that are informed.

SPEAKER_03 5:45

I mean, there's always that gulf, right? From research, and then you have to somehow make that leap into commercialization, into widespread adoption. Um and that can be hard.

SPEAKER_01 5:59

It is hard. It is hard. There is always uh, I would say, something I like to say is like we always need to understand what to draw a line between the real world and what the theory says, um to solve something that's actually been then possible to use.

SPEAKER_02 6:17

I find it absolutely astonishing when uh companies like Amazon, for instance. You know, I have two kids at home who are constantly on Amazon, they're ordering every day. So every day there's there's packages, but they come so quickly. Um I'm I'm absolutely amazed that you know you order at seven o'clock at night and then by lunchtime the next day, there it is. I mean, I think that that is it's it's so impressive, really, when you think about how far we've come. Um so can you I it's a sort of AI sounds quite scary. So, how how is AI being used in there? And is that one of the things that's benefiting um logistics now?

SPEAKER_01 6:54

Absolutely. It's uh the absolutely we are trying to use and leverage it um on a daily basis, I would say. And uh, as I mentioned, our lab was launched exactly with the idea of leveraging the power of AI. Um, I do understand, I do agree with the fact that it might sound scary. Um, and some somehow um makes sense if you if you think of AI as something that is meant to replace humans. Uh you don't want to you don't want to think that, okay, don't we are going to be replaced. The decision making is not going to be on our shoulders anymore. So what do we do? Um but the reality is that that's not what AI can do. Uh what AI can do is allow humans to um overpass the complexity barriers that supply chains of today um have. So if you think about it, every supply chain has thousands of SKUs, or most of the supply chains are large scale, thousands of SKUs, multiple suppliers, multiple warehouses. Um, a human is definitely able to judge, but is not able to judge such a big scale of a problem. Right. So that's where tools like Genesis um can support. So just to mention a couple of words about Genesis. Genesis is is a tool that we developed, is an is an optimization and simulation tool that we developed to improve um inventory management. So what it does, um it provides um support into decisions like which is the level of inventory for all my SKUs across all the warehouses in my network that I need to have. Uh, and when in time I have to replace to call my suppliers to ask for the replenishment of this inventory. Uh, it needs to do this basically um trying to balance all the trade-offs, which is cost, which is risk, uh, expectations. Um so this tool, as I was mentioning, is working with large-scale problems and it does it, but it also has a user interface. So we do not we did not only develop the solution, we develop the user interface, which is uh giving the possibility to teams that are technical or business to work with this tool. And usually business users have the entire understanding of the complexity and of the problem they want to solve. So now they have a tool that they can use to explore possible alternatives. Um now, to be honest, this is not new. Again, meaning um solutions that would support this type of decisions have always been there or have been there for many years. The difference that I I probably I see is, and there are two. The first one is because the supply chains have also always often be big, uh solutions like this were taking days, weeks to provide outcome. And um that was okay somehow, but it was also something that users were using just to validate what was a decision that they had probably already taken. Um what Genesis allows you to do is solve scenarios in minutes. And if you can solve scenarios in minutes, you do not only solve those that you want to validate, you solve scenarios that you want to explore. At the end, the decision is always on the on the on the users, on the people, on the on the decision makers.

SPEAKER_02 10:42

So AI is a great assisting tool. Yeah. Yes.

SPEAKER_03 10:46

And I think maybe if people looked at it not as, you said, not as a replacement, but as an expansion of what humans are already doing. And in a a really low margin industry, uh that speed of decision making becomes an incredible um competitive advantage, I think.

SPEAKER_01 11:12

Absolutely. But um it is an advantage, it is important, is it is what is needed today in a change in a world that is constantly constantly changing. Um, but if you think something again, going back to the importance of people that's never going to go away. Um, if you run different scenarios and you get different outcomes, it's still up to the human to understand if those outcomes make sense, if they are reliable, is there if they are reflecting the reality. Um, if the questions being asked to the tool are actually the right questions to get the right answers. Um, so it's important to have this possibility, and that's where we are trying to support with tools like Genesis. Um, but maybe that now, if we think at it this way, it sounds less scary to have AI around.

SPEAKER_03 12:04

And and you've really been talking about this amazing complexity that our supply chains have developed, right? With so many different, you know, many companies have different locations, um, you know, thousands of vendors, different inventory levels. And as I'm always telling my husband, you know, we look at the cost of something like buy it the two-pack of paper towels versus the 10 pack of paper towels. And I always tell my husband, well, you know, carrying inventory and storing inventory is a cost center. And he just looks at me like, what? But that is, you know, that is the type of on a much grander scale, that is the type of decisions that must be made accurately and considering all the data. And people can't always do that. I th I can see now, you know, as supply chains have just increased in complexity, I can see now how what an advantage it would be to have an AI assistant who can keep track of thousands of things at once or even millions of things. So, you know, speaking to that complexity, and I don't want to get too political, but right now global supply chains are being disrupted by conflict, by competition, by energy costs that are fluctuating, you know, so many things beyond the control of the companies that you're trying to help. So in paper and packaging, how can MILS adapt to these current challenges? And do you have a bead on like how long this type of insecurity might last in our supply chains?

SPEAKER_01 13:56

That's a that's a very nice uh question. And um it's it's fun because it makes me think of a presentation that I think I gave like five years ago. I it was not even at MIT, and it was about um disruptions in supply chain. And um, I remember me creating a slide saying disruption and uncertainty is becoming the new norm. Um, but it's actually it's actually the norm. Um, it's now becoming is the norm, and it needs to be, I think, treated as the baseline. I don't think we can think at something that is going to go away. Um and if I may add on this, to be honest, I also think that uncertainty in supply chains it's always been there. Uh, what has changed a lot, and you were touching this, Mark, earlier, uh, is the expectations um that we as consumers have, the speed that we expect. We want something outside our door the day after, today, in African. Um, this was not the case five years ago, eight years ago. We were like, okay, in ordering something that would arrive in two weeks, and we were happy about having it delivered, right? Um, now we don't even think of ordering something that's not there in two days. We get upset. Um, so I think this is playing a huge, huge role on the fact that we see uncertainty today more than we were seeing it in the past. Um if I think at if I think at it from this perspective, um for me, it's for any type of industry. Um, I I don't think the question is how and when things will get stable. It's more about how do we deal with something that's going to be always unstable. Some approaches are, you know, like in the past, supply chains were operated uh with the approach of making them efficient from a cost perspective. So uh having the just in time approach. So let's get things into the warehouses just in time. Let's not have an inventory that is too high uh because that saves and costs. Going back to your inventory question, it it was an approach, it was approach. Why not just in time? We're fast, we get things. We can uh we can have this philosophy of reducing our inventory, saving on cost, and get things only when we need those things. Um, now probably this needs to be changed, and actually it not probably that should be changed. We should move more into a just in case type of idea. Yes. Just in case something happened. Yeah, yeah.

SPEAKER_02 16:52

Yeah, so that you've got something there in an emergency situation.

SPEAKER_01 16:56

Yes, just in case my supplier fails, just in case uh the delivery is late, uh, just in case something happened and I have only one supplier, but it's probably not a good idea. So let's diversify the number of suppliers that we have, uh, so that if one is late or fails, we can count on another one. Um and regionalization is something that several companies are are are trying to um work on, trying to rely also on things that are closer.

SPEAKER_03 17:28

And that's where a tool like Genesis really comes into play, because it can keep track of where everything is, how quickly it can move from one place to another in a way that I certainly wouldn't be able to keep in my head. Um with interface that's yeah, no one would, right?

SPEAKER_01 17:50

No one could. Uh no one could. Um if I can add on on this, is um tools like Genesis would allow you to also don't think that the decision you make today is gonna last for years. You need to revise that decision.

SPEAKER_03 18:06

Yeah, yeah, and it can run those scenarios for you of the different decisions you might be able to make.

SPEAKER_02 18:13

Yeah, absolutely. Like a digital twin sort of thing, yes?

SPEAKER_01 18:17

It's uh yeah, it's it's more of a like a supporting tool that you use when you feel you need, and you probably today need it quite often, um, to get possible uh possible different choices that um come from changes in demand, come from uh a failure in supplier or any other unpredictable thing uh that you want to analyze and be ready if there is the need to change. Let's navigate the uncertainty uh more than try to wait for it to get to an end.

SPEAKER_02 18:53

Yeah, that's so true, actually. When you look at it, we've been through so many different challenges over the last six or seven years. You know, we had COVID, then wars, you know, then more wars, uh then oil crises, and and it's fluid, isn't it? It's constantly fluid. It's always one disaster going on somewhere else and another one appearing somewhere else. So you've got to have your logistics so that they're also fluid and they can adjust accordingly.

SPEAKER_01 19:19

Yes. In Italy, we used to have on the on the windows in our houses that everything would get better soon during COVID, right? It's gonna end. COVID ended, we are we are happy enjoying, but there are other challenges that are coming up. It's not um it's we need to navigate into it, I guess. Um that's the the idea.

SPEAKER_03 19:41

Do you think that's only important for larger companies, or could smaller companies also benefit from that kind of assistance?

SPEAKER_01 19:49

No, I think I think that that's that's important for any for any company because it I um maybe the scale can get smaller if the company is smaller, but it's never going to be just decisions that you can count on your fingers. Um the complexity of the multiple decisions that need to be taken in parallel is going to grow like quite soon. Uh, you don't need a large scale to get to several number of possible uh scenarios that you might want to run and the decisions that need to be handled um in parallel. So I would say no.

SPEAKER_02 20:30

Great. You've got a fascinating job. That's no, there's no doubt about that. So can you tell us what's ahead for the intelligence logistics systems lab? And also how can paper and packaging manufacturers stay in the loop um as far as new things are developing?

SPEAKER_01 20:46

Yes. So I mean, probably I I I hope I made it clear from the very beginning. We are really um intrigued by new challenges. Uh so what we will continue doing is try to always work um in partnership with our like the partners company of our center with the idea of getting new challenges from them and use um AI data operational research to help um those our partner companies to face those challenges and get a solution that helps them um take better decisions. Um, so that's I would say that's where we are going to continue with working. Um for paper and packaging manufacturer, if they want to be in the loop, we are very active on LinkedIn, for example. So we always publish things uh that we do um on such platform. Uh and it could be like following the center itself, our lab, individual people working at the lab. Um, we do have a mail, um, sorry, a newsletter that um They could subscribe on on our website or just reach out. We love to hear about challenges. So usually that's how conversations start. So why not?

SPEAKER_02 22:13

Sure. What's the most interesting challenge that you've had in the in the sort of um paper industry? Is there one that you have?

SPEAKER_01 22:20

What I something I did um the first year I joined, I joined actually CTL. We work also with students. And um that year, like when I started, my my my boss came and said, Serena, do you have an idea of a research of a capstone project for students that were starting the capstone um their master that year they could work on? And because I just moved, um, I was basically seeing myself on a daily basis get delivery because I was furnishing my apartment and going downstairs and upstairs with all these boxes where you could not even see me. New, super new boxes, super new. They were really just coming from a warehouse, get to my place. I would think, I would take things from inside, furnish my apartment, and then throw them. And that was for me, it was very tough to see. Uh, so I was like, how much of this is actually going to be really recycled? Uh, how much of this is going back? Uh, and and someone else will benefit from it as well. Um, so that that, you know, like it was it was really nice to do the journey with with the students, but everything really started with like, okay, these boxes are so new, and I'm just throwing them right away.

SPEAKER_03 23:37

So, so Dr. Sylvestria, are you telling me that this MIT students capstone project was helping you carry boxes up the stairs? It was because I think that's great. I think you deserve that assistance.

SPEAKER_01 23:56

No, I seen that three floors to climb, no elevator. So uh, but at least I was, you know, like thinking when I was putting them in the trash, I was like, okay, they're gonna be recycled and actually be used again, uh, which is nice. It's a nice feeling. I really care about it.

SPEAKER_02 24:12

So it was yeah, but it seems so crazy throwing something so new away, doesn't it? It's it's it's really yeah, it's so crazy.

SPEAKER_01 24:19

So could we reuse them and return them as they are? Uh that was actually one of the questions of the students. Can we actually uh as we drop new deliveries, can we actually pick the empty boxes back uh as they are? That was my open question.

SPEAKER_03 24:37

Yeah, I think that's a a fantastic model, you know, and maybe something that will be there in the future, right? Where you have a little bit more of that uh that manufacturer gets them directly back, right?

SPEAKER_01 24:52

Yeah, absolutely. The the study was like, of course, you need to think that to collect back those empty boxes, there is transportation costs to consider. So it's a balance. Uh, but if you think about it, I'm sure someone in my building, if not me, will order the day after for something else. So someone will stop by to deliver, why not to pick up what was empty and left out there? Um, so maybe optimizing, you know, like the can we merge the delivery of something new with the pickup of something that is there um and is being used just for half an hour?

SPEAKER_02 25:28

That was a great answer because I threw that into surprise you.

SPEAKER_03 25:31

Yeah, you threw it into surprise. Well, just speaking of speaking of surprise questions, are you ready, Mark? We'll we'll throw some lightning questions at you.

SPEAKER_02 25:41

I'm ready. What's the biggest change you've seen in the packaging industry over the last 10 years?

SPEAKER_01 25:51

Uh the need, like with this change on speed and complexity. Um, yeah, that you just see a lot of it. Like you, I think you're facing more and more the awareness that there is a lot of it being used every single day for um the expectation of customers to get things delivered at their front door before COVID, we were not used to it, right? No, big changes. Yes.

SPEAKER_03 26:18

All right, I have one too. So we've talked about the complexity of supply chains, right? What is one thing that you wish uh everyone understood about managing supply chains?

SPEAKER_01 26:34

That there is on there is never going to be only one single optimal decision. Um and um yeah, the the most important thing is evaluating different scenarios and taking decisions that balance things because there is not one single decision.

SPEAKER_03 26:52

I love that answer. It takes off a little of the pressure, doesn't it? Because it's so complex. I think one thing that people I think one thing that slows adoption is that people are afraid of making mistakes.

SPEAKER_01 27:08

Yes, um, understandable that um I think we should we should live with that. There is so much uncertainty that mistakes should be part of it.

SPEAKER_02 27:20

Yeah. No, brilliant answers. Absolutely brilliant. Yeah, really, really good.

SPEAKER_03 27:27

I agree. We've loved having you on the podcast, Dr. Sylvestri. Thank you so much for your time today. Thanks for you. I really enjoyed it. Thank you.

SPEAKER_02 27:36

Cool. We must meet you again at TappyCon and get you coming up and giving us a speech. That'd be awesome. We are so glad you could join us for this episode of Uptime. You can learn more about Tappy Publications at Tappy.org. I know you'll find something that's perfect for your career growth.

SPEAKER_03 27:57

Please subscribe to the podcast channel and share this episode with your industry colleagues and friends. Because everyone could use a little more uptime. Thanks for being here.

SPEAKER_00 28:11

Hi, I'm Lawton Roberts, president and CEO of Tappy, the pulp, paper, packaging, tissue, and converting industry's trusted resource for the publications, training standards, and networking that are so vital to our industry's growth. Tappy conferences are an excellent way to build both your knowledge and those lifelong relationships that are so critical to your career. When you come to our live events, please stop me and introduce yourself. I would love to hear your story. In the meantime, check out Tappy.org for more. Thank you for listening.