 Welcome to the SC4X second live event. My name is David Corell. I'm your course lead. It's been a great pleasure to work with you up to and including this week's midterm. I hope it went well for you. The scores are looking excellent. Thank you for all the time you've given to our course. And I'm very excited about today's event, especially because of our guest. I'll ask you to introduce yourself here. Please tell our learners where you're coming from and a little bit about yourself. Yeah. Hey, everybody. Thanks for taking the time to tune in. My name's Harris Ligon, and I represent Uber Freight. I'm currently based out of the Chicago office. We have a couple of geographic locations, mainly in San Francisco and Chicago and some other folks doing some field work out in some various locations. I'm representing the team that really focuses on designing solutions for our customers on the demand side. So when we think about Uber Freight and we think about the broader world of transportation, everything can be broken down to both supply and demand, supply we think of from our driver and carrier base and demand we think of for folks that need to move things around the network. And so I spend a lot of time with an amazing team thinking about differentiated solutions that we could bring to bear that help enhance their access to the market, help their understanding of the market, and hopefully help them transact and execute more effectively. Excellent. Thank you so much. I see a few of you are already waiting in on Slido. So we're going to get to all your questions. Remember that you can put those in and we'll ask them to Harris. I'm going to start ahead on some of our agenda items here. OK, just making sure everyone is getting the broadcast. All right, so welcome to the Second Live Amendment. We're going to talk a little bit about Harris and his career path. And hopefully that will be interesting to you and maybe inspiring as you think about doing some of the same kind of work. We're really going to be interested in your opinions and your thoughts. I see a note already from one of our viewers. We'll try to speak a little louder. Thank you, Mustafa. And then I'm also going to talk a little bit about your next procedural sort of course headache, which is the proctor graded assignment. But let's have fun before we get into that. I think one of the things that sort of motivates where we're going is to think about what do we need data analysis for in transportation? If I make it at its most simple point, I think I'm going from A to B. I could understand needing a horse or a car or an airplane. But we're spending a lot of time in this course thinking about data and how we can use it. So one of the first things I'd like to do is sort of pose this question to Harris and just think about what are some basic ways that people use data in transportation? Yeah, so I think it's always easy to think about the world from a procurement standpoint, because that's something that I think a lot of your learners and myself being involved in the program have thought about is really understanding what are the decision variables that are going to drive me towards wanting to make A decision or B decision. And for transportation, data is really important from a point A to point B perspective. Because if you really think about it, there's always a profile of what your decision is really going to be when it comes down to like, David, are you going to buy this thing? Are you going to buy that thing? And if you do, what's really driving that extension of that transit? And so at Uber Freight, we think a lot about what are the variables relative to the market. So when I think about it procuring a thing off of Amazon or thinking about procuring a truck, I'm thinking about how much is it going to cost and what is the market telling me what that cost should be relative to a low profile. So time of day, weight, length of haul, appointment times, all of those things are variables that we think about when it comes into transportation. And those things are important because we think about some earlier classes. It's important to know those data points so you can actually forecast effectively. Because when you think about it, there's a lot of people moving a lot of things across the globe on a day-to-day basis. And without those data points, they're not going to be able to really understand what their projected spend is going to be and what their variance is relative to that projection. And so when you think about spending money and not only for that procurement but investing capital so that you can produce items or goods or you can actually think about delivering on those items for people who are procuring those, it's important to have a good understanding of that so you can think about that in the future and forecast that relative to what you think it needs to be. And so from a data analysis standpoint, you're extracting knowledge and understanding from all of those things and lacking data. It's just kind of a bunch of ad hoc decisions that you hope add up to a good outcome for your customers. Thank you for that. It seemed like it really confirms what some of you have said here in your responses. So if you're on Slido, you can see we've got the word cloud of what you're talking about. We're talking about cost, decisions, supply, mode. It looks to me like that's a mode selection and network design. I think what was especially interesting, you were thinking about forecasting. Could you talk a little bit about what are we forecasting for most importantly when we think about forecasting and freight transportation? Yeah, so I'll actually cut this in two ways. So my background prior to coming to Uber freight was not in trucking, I was a rail guy. I did a lot of thinking about the railroad industry and marketing and selling those products. And so one way you could think about data and making a decision is what mode is that ultimately going to take? And so you obviously have, there's potentially an ocean or an air component when you think about surface transportation you have a variety of things but mainly that evolves in truck or rail. And so data is important from having, being able to optimize for a route guide or ultimately a mode selection very much upstream and kind of the decision process. And then later on when you think about being able to forecast that you're gonna have some constant puts that they go along with that. So generally speaking, rail is often cheaper than truck and being able to forecast that based on optimally good decisions based on load profile and length of haul and transit schedule, being able to forecast those things early is gonna help you determine what the best cost is ultimately for either A, for you to deliver that service for your customers or you to minimize costs for your own organization. So you either want, really it comes down to maximizing profit or minimizing costs. Perfect, thank you. And thank you all for your many responses. This is great. I think where I wanna jump now is to just ask Harris to introduce himself and tell us a little bit about how we got to where he is. He just sort of started us on that path. We learned that you started in rail and now you're working in some really innovative stuff with trucks. Could you tell us a little bit about your educational background and your professional background? Yeah, so I'm actually, I think I'm kind of a non-traditional person both for the trucking industry and for the tech sector. Cause normally coming from a rail background, that's an area that isn't normally thought to be very tech forward and tech focused. And the truck space is typically something that competed heavily with rail and continues to do so, even in especially today's marketplace. So my journey to Uber freight was really born on the fact that number one, there are tons of opportunities to recognize inefficiencies whether either in your own personal life, maybe where you work or within the industry that you're working in. You get to see those every day. There often are not exceptional opportunities to be able to go forward and address those very effectively. And Uber freight for me was that opportunity. So I was able to make a transition over to that organization in early January of 2017. And I've been able to do quite a variety of things there from really just building a business from day one which has been really exciting. My background, I started out in undergrad in psychology which candidly, as you and I were kind of talking, that's really helped me understand where having empathy from where users are coming from, from customers are coming from and then broadly across the organization helping myself understand what they really need to solve problems. Because that really is kind of the business of supply chain. You're just solving problems. And so from there, I spent some time working on an MBA at the College of William & Mary and that was a really enjoyable experience. I always encourage people whether you're coming from a more liberal arts education or a more engineering quantitative focused education that if you wanna stay in business, a broader business education can help you understand that the problems either you know from a humanistic perspective or from an engineering perspective, they have some real application from generally like just building a business that ultimately can sustainably solve problems for folks. And then I've been working through a couple of courses on the edX platform and I really find SC0X and SC1X and working through SC2X now. That has been really engaging for me because I'm able to take forward that education that we're working through on a weekly basis and actually apply some of those techniques to what I'm doing on a day-to-day basis. So being able to think about network problems and optimization and vehicle routing problems, those things are obviously super applicable for Uber Freight but I think more often than not being able to take those techniques and skills and be able to place those in the pockets of our customers. That's been really beneficial as well because we feel like we're really doing something with that knowledge. So yeah, I think I'm really, really happy with what the MicroMasters program is doing and I encourage everybody who's watching whether you're just starting out or you're thinking about starting out or whether you're wrapping up SC4X to be proud of what you're doing and to just take the leap. Oh, thank you so much. And we're so excited to have learners like you in our courses who are doing great things of practice and giving some time at the end of the day to our program as well. Just because I'm always curious how our learners do this and learners who are watching, be in touch with me on email and let me know. Where do you find time to do the courses? Do you do it at night? Do you do it during your lunch break? When did you sort of cut out those hours you needed to watch the videos and do the work? Yeah, if my team was watching this, they would laugh because they know I'm a very non-traditional person. I wake up super early and oftentimes that would be where I would get my repetitions on the videos and some of the quick questions. When it came to some of the heavier lifting like the practice problems and prepping for the homework, a lot of that time was spent on the weekends and so you're just carving out time. And so the best advice I could give to anybody is have a calendar, use it and just block out the time for those courses in your coursework and just stick to it. If your calendar is always kind of your guide, you'll be able to figure out time to execute all these. One of the things that impresses everyone here in the MicroMasters team about students like yourself and our learners is that discipline to get this done while doing other big things, committing to the calendar, the schedule, making the sacrifices. So much respect, sir. Thank you. I wanted to follow up on one thing because I thought it was interesting. When you were studying psychology, at that time, were you thinking about it in sort of a business context or did that come later? It actually, so for me, it actually came later. I really, so my kind of like my, the point that I work from always as a leader is that it's people first mission always. And so really focusing on people first for me was that was kind of the area where I was working from and the kind of the mission or more the applicability came later for me. And I think that's one of the really cool things about supply chain. I referenced this early, but you're just in the business of solving problems. And oftentimes within, especially in transportation logistics, you get to solve a lot of problems very quickly. And so if you'd like to make decisions, if you'd like to execute, if you'd like to actually see the needle move, it's a great place today. Yeah, yeah, yeah. And do you think, I'm getting a little bit philosophical here, but I think it will be interesting. Do you think the background psychology has helped you at all to understand or to think about any of the quantitative tools? And I guess I'm thinking especially about the machine learning. Is it, did they overlap at all or do you seem as different? And I suppose just to sort of set you up why I say that is my colleague who some of you may know, Dr. Friedman or you may have seen Lex in some of our videos. He gave a very interesting interview last week and he was talking about, he got into artificial intelligence because at first he just wanted to understand the human brain and artificial intelligence is just the beginning efforts of rebuilding a human brain to see how it works. Right. Yeah, so I think to your question, do I see an overlap of psychology with AI or do I see that application in logistics and transportation? Absolutely I do. I have two folks on my team that really do some amazing work. Amaro and Christina and it's funny they both commented at separate times that when they're working on big algorithms or they're doing big data sets and with analysis and leveraging some machine learning techniques that really they're just kind of setting up a decision tree which is really is not unlike what we would be doing where we're kind of like you and I are just like working through a problem right now like we had to go solve something. And so that kind of digitalization of the process, I think definitely lends itself to having like a good sense of psychology and understanding what like what incentives are and what drives like the final decision, right? Yeah, that's really interesting. Gosh, I'm gonna jump into just some of the questions here. Thank you for putting these in. It looks like we have a great audience. Chris asked, and this is probably an appropriate time. Could you give a quick overview of the business model of Uber Freight? It'll probably give context to some of these answers. Yeah, so obviously this is a global audience. So I understand if not everybody knows about Uber, right? But effectively Uber Freight and Uber more broadly, they are interjecting a technology platform between the demand, which oftentimes is a person who needs to move, a good food delivery that needs to happen or a shipment that needs to move somewhere across the network. Then there is a supply of oftentimes couriers for drivers or truck carriers that will be moving those things. And so the beautiful thing about Uber and Uber Freight is there, we've created this beautiful platform that just allows things to sync up. And in the background, you've got AI running and machine learning that's determining what kind of an effective market rate will be in order to incentivize the procure of that transaction to execute on it. And so really the basis of Uber Freight is saying there are a bunch of people that need to move things. And oftentimes those are consumer packaged goods, those could be industrial products, a variety of things that need to move in any truck load, whether that's refrigerated or simply like over the road trucking. And then there are carriers that their entire business model is predicated on being able to move those things. What Uber Freight is trying to do is to accelerate the velocity of those transactions and build up the density on both sides of the marketplace. And if we do that, we do that really well, we create kind of this virtuous cycle of liquidity that transactions are just happening as fast as you need them to happen and they're happening at a rate that is fair for the market at that time. That's great. And maybe now would be a good time to talk about the power loop and the things that we were talking about there. Yeah, yeah. So just launched real recently is a new asset program that Uber Freight has kind of gotten access to. And so traditionally for those who don't know, traditionally in the North American truck market, if David and I decide we're gonna go start up our own trucking business, we'll both go procure tractors or power units. And then in order to haul freight, we actually have to go buy trailers which sit behind the power unit for us to be able to navigate and move things. And that's how we generate revenue and we hope to minimize our costs in doing so. The problem is, is both of those transactions are pretty costly. And so we've partnered with another organization that allows us to effectively create a nationwide pool of trailers and assets so that individual truck owners and operators don't have to go procure the trailer in order to be able to move things. They can simply opt into this program, they're thoroughly vetted and we do all those things in the backend. The cool thing about it is they simply can book a transaction on our platform and then they just show up and move that. They don't have to carry a trailer with them. And then when they back into a facility, they're not waiting for that transaction or that loading to happen. They simply show up, grab a trailer and depart. And then when they get to the destination more often than not because these trailer pools are being scaled up nationally, they're able to simply just drop off that trailer and move on to either another trailer that's waiting on them right there or they can just go about their business for the rest of the day. And so the cool thing about this is that more often than not drivers waste an inordinate amount of time waiting for a truck to be loaded or unloaded. And those hours obviously have a cost with them but they also have a cost from the fact that the supply chain all of a sudden gets longer because you have a bunch of people that are involved in that value chain that are just waiting around for something to happen. And so by bringing power loop into there it unlocks the opportunity for Uber Freight to offer power only options for the carrier base. And a couple of things happen as a result of that. We reduce the cost of capital for being able to get into trucking. So we actually lower the barrier to entry. The second thing for shippers on the platform that opt into this program as well, we actually reduce congestion at facilities. We actually accelerate loading times and we improve their focus which really just should be on warehouse management and we help them get better utilization out of their dock doors at those facilities. So we see benefits on both sides. It's a fairly new program. We're still in the early stages but we've got a great group of people that are working on it and they love to think about network problems. They love to think about like, it's a classic vehicle routing problem, right? So you've got this individual asset that you're trying to keep in this constricted space and how do you route these things around? And fortunately it's one of the things that we've partnered with MIT on in the latest round of. Yeah, yeah. When it was really exciting too from a research perspective because one of the first conversations you and I had last time we were here was thinking about, so there's a shortage of truck drivers and so we're trying to move things around with less and less drivers but sometimes we can think about not just the number of drivers but the hours available for driving and it sounds like you're moving people through faster with the system. Right, so ultimately at the end of the day a carrier should be able to earn more because they're actually actively moving around the network and completing more transactions throughout their tour of duty. So whether that's more transactions per day or more transactions per week, however you want to measure it ultimately at the end of the day their earnings should be able to go up and their access to better freight should go up as well. And potentially if you think about better earnings and a lower barrier to entry that's another way to address the drivers. Excellent thing. Gosh, so many good questions coming in. I'm gonna jump ahead to some of the stuff we had on our agenda quick but I think we're gonna have time for you all to do what I think is a lot of fun is just picking Harris's brain about interesting topics. Let me start here and then go back. So one of the things that we talked about is sort of the interface between managers and AI in supply chain and logistics. And I wanna take this opportunity to give a big shout out to our community teaching assistant Alessandro who really went back and forth with almost every learner who posted on this. So big hello and big thank you Alessandro. And I wanted to talk to Harris. What do you think about the interaction between AI and humans is are we gonna be replaced? Do we need managers in the future if AI keeps moving forward? Where do you see it going? Yeah, I think that AI will always augment what humans are able to do. So I think that there will be a subset of manual tasks that are occurring right now that will hopefully just be automated and their efficacy and completion rate will be somewhere like the error rate will be so low that we won't mind having AI taking over those manual tasks. And what that does is that actually frees up humans to take on more critical thinking, more higher focus, maybe more strategic decision making. And candidly, we need more people making those decisions and thinking about modeling out decision trees or thinking through whether this is the right decision for my business, right? Because the model will give you an answer. And that's really cool. And leveraging AI to kind of evaluate a very unstructured data set or like even photographs, for example, right? There are tons of really cool applications for that. But ultimately at the end of the day, the recommendation may not be right for the business. And that's ultimately where humans are gonna have to be. And so at multiple levels, I'm really excited that AI will augment what humans are able to do. But ultimately at the end of the day, humans will still own the core decision making around what that strategy will look like. Let me go to the polls here. Why don't you answer, what do you think the role of humans in supply chain is, let us know and we'll look at your insights as they come in. Another interesting question that you all debated and I wanted to give a big thank you to Ed Marr who found this article for us and really inspired us to think on this route. Is AI better for revenue generation or cost generation? And why do you think it's better for one or the other? Learners, as you address this in the forum, you sort of sided with the authors to say that it was more of a revenue generation thing. But let's talk to Harris and see what he thinks. Yeah, I'll frame this up from the perspective that I think AI has two really, really good applications. One of those, and we all know about this, this is how I think we naturally think about AI, it's prediction, right? It's gonna tell us what will happen at some point in the future. So from a consumer standpoint, we could apply AI and really extract some meaning from oftentimes unstructured data sets. I think even Airbnb uses this to think about if a person is spending a lot of time on their pages, or they seem to be clicking through some photos, oftentimes maybe they're making decisions on those photos, not necessarily what is in the super structured data set that's over here at the left that tells them how many beds, how many baths, or those fundamental things. They're spending more time thinking about what their experience is gonna look like. And so AI can oftentimes predict what a consumer decision may look like. And that's definitely how I think about that. I was just a little quiet. Okay, sounds good. You were golden. And then there's the other side, which is, so we talked about prediction, the other side of that is detection. So understanding what's actually happening, because when we talk about augmenting human functions, means you're not having people that are maybe checking a lot of invoices to make sure things are lining up, or making sure that the fundamental digital operation is running optimally, right? You never really wanna assume that humans are gonna be able to do that really well at scale. So for us, when we think about the number of truckloads that are moving on our network at any given point, as that number goes up, our ability to really manage those things on an individual basis becomes smaller. So we need to manage the exception. So AI can also detect when things are not running really well. So when I think about both of those opportunities, I am of the belief that I think AI can be actually used for both. So I think absolutely it can drive up revenue. And I think that is thumbs up on that. I'm a big ref, big guy. But I actually think it can reduce costs. And so when you think about a supply chain, if your AI is evaluating, let's just say like your very small supply chain that may originate in India and have maybe some air transit and some ocean transit. And then once it lands, maybe in LA Long Beach, it's gonna move maybe by a rail or by a truck to some other point in the US or maybe in the UK, however it plays out. Looking for exceptions in that supply chain when things are out of alignment? That's potentially a cost savings measure as well. And so I think AI has application for both. Excellent, thank you. Let's take a second to look at some of your answers and see what Harris thinks. So when we asked you, what do you think the role of humans is in supply chain? I like this one. Just like any model would do anything for a dollar. Humans have to be there and assure that decisions will not be taken just because of that. Yeah, I think like when you really think through that, yeah, I mean, in some ways, it's have to be the checks and the balances for what AI will eventually become. And I think that's kind of one of the most interesting things about AI for me is that we all have this opinion or this assumption about what it will or could be. And that is just limited on our knowledge based on what we know today. And so I think like thinking about AI in the future, I think it's really good. I think that humans will always play a role of validating and proving decisions and moderating. You're right. I'm seeing decision-making a lot of problems. I fundamentally agree with that over and over. Let me change the question now to you all. I'm gonna put forward the role of AI in the supply chain and ask you to come up with some specific examples where AI could be used in the supply chain. And while you're thinking of yours, I might ask Harris to follow up on something he and I were talking about a little before and you alluded to just now when thinking about prediction and you were talking about the trailers and setting up the load for the drivers, is there a role for AI there for finding the right load for the right driver? Yeah, there absolutely is. And so I'll frame this up in a couple of ways. I think the first way to think about this is just from a pure operational standpoint. So let's say Dave and I are running a team where we're driving a truck around and ultimately we may select a load that maybe moves from San Antonio, Texas to Houston, Texas or maybe Madrid to Barcelona. Once we make that selection, that's great for that initial transaction but what happens when we get to our destination? And I think that's where the role of AI and prediction really comes into play because ultimately we should be able to predict based on the driver's preferences or maybe their previous history or a variety of other things what a good potential transaction looks like for them in the future. So Dave and I get to wherever we're going. Hopefully we can already have our next set of loads already picked out. And that's one of the things we do with UberPay. We have a feature called reloads and it's really predicated on this idea that we can bundle similar transactions together or chain transactions together much in the same way that we would on a vehicle routing problem based on a user of preferences. And the cool thing about that is like that is an active utilization of AI that is going on right now thousands of times across our network every hour. So a driver will open up a shipment they'll be evaluating it and they'll also look at the bottom of their screen and they'll see very clearly well when you get to your destination based on the hours in the load profile and how long it may take to unload here are some other options that maybe makes this initial transaction much more enticing. And from an optimal standpoint so we just now optimized for the operator based on their preferences in previous history. But if you think about it we're helping out the demand side of the network as well because the more often you can keep an asset utilized and moving around better it is for everybody. Absolutely and I think for our learners I mean, could I understand that correctly if I were to think about it like the way Amazon recommends things for me to purchase based on my press purchases or Netflix recommends things for me to watch are you recommending loads sort of like that? Yeah, Amazon's got a very interesting model and they are trying to take money out of your pocket data. What we are trying to do. And they've succeeded many times. What we are trying to do is to make sure that drivers have the best earning opportunity based on whatever their preferences are. So that's what we're recommending to them based on whatever demand we have in our network the best way they can match their supply. I see, yeah, yeah. I guess thinking about for our learners when we think about and this is sort of an invitation to you all as you're thinking about the final and working with the material we talked about association learning. So these things are like this and it seems like there's perhaps some sort of application there. And the other thing I wonder if it's part of it is we think about cluster analysis and we think about that sometimes when we think about transportation would you group different sort of parts of the country as clusters so when the driver goes into that area you're looking in that area for the right load for them? Yeah, I think that's definitely a future like that's definitely an application of AI and especially like when you zoom into the supply chain clustering has great benefits from kind of an optimization and a structuring perspective, right? You're optimizing, you kind of zoom out a little bit or you're optimizing for fewer points because you're actually grouping them together. The downside of that, especially when you get when you get into kind of final mile execution or loading and pickup is that those clusters may look really good, but oftentimes they have variables for each of the origins and destinations that I think you actually have to account for. So clustering makes logical sense for me especially if I'm trying to like run a super optimal network. I would want a cluster from thinking about it from a driver experience standpoint making sure that we make sure the market rate is fair and equitable for the different points on the map. I think that's a challenge that we definitely are evaluating. Thank you. Let's look at some of your thoughts here. What is the role of AI in the supply chain? I think Hasmat absolutely for those of you who have experience working with Hasmat, it is very nuanced and very detailed portion of the supply chain to be able to get those things right because obviously if you don't, there are a variety of consequences as you remember that. The AI definitely can play a role there in making sure that pushing well up the supply chain that the right information is attached to the right load or right transaction. And when I think about like somebody just posted empty leg fill, 100% absolutely. I mean, there's a, I think there was a statistic at some point and this is anecdotal, I don't have a source right now but I think like 20 to 30 to maybe 40% of miles driven by trucks in North America are empty and that's a lot of waste going on. And so one of the things that the team that I lead is able to do, they're able to kind of match a customer's network and thinking through, okay, like here's what they move on an annual basis and here's where our carrier base will be at any given point based on previous history and a variety of other things. So I think there's plenty of opportunity out there to apply AI to real live, hard to solve challenges within supply chain. Can you explain more? It sounds like you and the learner, we don't have the name, but there are a couple actually mentioned of Hasmat here. Is it AI as sort of a information check on sort of the things have been stamped the way they need to be stamped or secured the way they need to be secured or is there an optimization component to AI and Hasmat transportation? Yeah, so as I'm walking through this a little more in my mind, I think there is an optimization standpoint when we think about it. So at the point of demand signal and knowing that that's gonna be associated with a Hasmat shipment or Hasmat transaction, filtering that down and making sure that there are the right type of operations or security features or just the right information to make sure that that load flows or that transaction flows in the most effective way possible because there's nothing more frustrating when a customer wants to move something that has a certain requirement and that requirement isn't necessarily able to be fulfilled. And so understanding that on certain days of the week, let's say for example, one of the things that I always thought was just amazing about Uber from a personal standpoint was that you were able to book Ubers that had car seats in them for small children. You can do that? Yeah, I did not know that. So as a dad, I remember thinking about that as an option and we can only do it in select markets, right? We don't have enough supply and enough demand to be able to make those transactions work. But understanding when a Uber partner needed to be on the network and they needed to have a certain piece of equipment in their car, that's absolutely mind blowing, right? And so you can fast forward that to other realms of the supply chain, but ultimately at the end of the day, as a platform, we have to be able to incentivize the right things at the right time. There are tons of applications for that. And that's a really interesting way to turn something on its head that we think about a lot sort of in my laboratory and with the on-campus students is preventative maintenance. So trying to predict when something's gonna happen based on the data and we can think about machine learning applications of that to try to find out where failures are. But you've sort of turned that on its head and said, where might we predict these unique opportunities like the class ban and stage for it? That's really interesting. And so that kind of going back to the initial prompt here, like that's where there are really opportunities for both cost reduction and revenue generation. Yeah, yeah, oh, interesting. You know, it reminds me of a question that I think our dear learner and CTA Param asked, is there a role for the internet of things and blockchains in what we were just discussing and thinking about sort of labeling load as hazmat and following all of the paperwork through? Do you think there's a potential there? And is Uber seizing that now? Yeah, so to answer the question quickly, yes, I think there's plenty of opportunity. I think Uber is always evaluating the application of a variety of technologies to our platform and being able to deliver value for both people who are partnering with Uber, but also are using Uber to procure some sort of service. So I think there's plenty of opportunity out there. I wouldn't say that we're in the throes of blockchain, but I can definitely say, I can always tell you that Uber is always evaluating opportunities to make our services more effective. Going back to the internet of things, one of the cool things going back to the power loop item that we were discussing earlier, we definitely have tracking live GPS real stuff going on with those associated trailers. And so that's not only is that a value added service for folks who are partnering with us on the power loop program, but it also allows us to kind of have a better understanding in real time what's going on with our assets. And so leveraging that information in addition to real-time cell phone tracking helps us really understand kind of like how we need to model and build the network more effectively in the future. And actually we're right down the street here, when we think about internet of things, what we're on the topic, there's an organization here in Cambridge called TIVE. I don't know if you've chatted with them recently, but they've come up with a pretty small sensor that is both GPS and cellular enabled. You basically just stick it right on something and it'll track it pretty effectively. And so for those of you out there that are thinking about the variety of applications that you've learned about in a multitude of courses, there are organizations out there that are working on some really cool and innovative things. And if IoT is your thing, like there's definitely opportunities out there. Thank you, that's a good one. Josh, in thinking about that, I know from working with the carrier community that sort of expectation now for real-time location and visibility is something that people are trying to adapt. And I didn't think about enabling that in the trailer as a sort of a quick solution for the smaller owner operator, that's interesting. So it's an investment, they don't have to make themselves, right? Yeah, yeah, yeah. So you know what we think about maximizing revenue and minimizing costs, right? Like there are tons of solutions that technology allow for us. And if done really well, you can take that data set in and predict from it or detect from it. Yeah, yeah, yeah. You know, I think that follows right nicely into a question. So that's a huge data set. I mean, if you're working with a file that's gonna come off of that kind of thing, what kind of advice do you have for our learners who are people who are interested in transportation and supply chain? Many of you are like Harris already in careers that you enjoy or in the industry, but for those among us who are thinking about that next position that is going to be sort of innovative and data intensive and doing some of the things that you're doing, what are the tools? What are the skills? What are the packages that you think people should be comfortable with to work in this world? Right, well, I'll tell everybody that that's online. You're already making a really good decision by investing your time and energy into this program. So I think that's step one right there. I think there are a variety of tools and skill sets and fundamental knowledge is like understanding SQL and being able to use SAS for modeling or optimization and really like being able to articulate that maybe even in just like an Excel spreadsheet and understanding how those two problems can be solved in multiple different fashions because oftentimes you're not gonna have access to both tools. But I think really fundamentally like the thing that I think everyone should take away is knowing the methods and getting to an end result is great. And that's absolutely, we should all be solving and striving to be able to come up with the right answer. The challenge, and more often than not, this is the challenge across every organization is being able to sell your optimal solution upward and inward within the organization. So oftentimes that leads to having some healthy conversations with folks that the solution in your model and your method really makes sense for the business. And sometimes that means you're selling it externally as well and having some good conversations with customers as well. And so those behavioral skills are equally as important as being able to come up with the solution, right? So I think spending time with your supervisors, with your teams, with digging deep and I fundamentally lead with questions. And so oftentimes when I get into a room with someone and we're walking through an analysis, how oftentimes you've leveraged two phrases, tell me more or walk me through. So I really wanna dig and unpack and understand. And so if you're on the flip side of that conversation and somebody's asking you those items, get comfortable with being able to articulate your decision, your method, your methodology and distill it in a way that just makes it a yes or no decision. Because ultimately when you drive, when you're having a conversation with someone and you're explaining those things, getting lost in the minutiae of your analysis, that's not a winning position. The winning position is ultimately driving down to making it a binary choice, zero or one. Are they gonna say yes or no? And the more simple you make it, the more articulate you are and the more straightforward you can be with the outcome, you're more likely than not to give a yes. That's great, thank you. And it really reminds me of, for those of you that went on the virtual field trip to New York and you met our friend, Michael Kress from ABI who told us that the project doesn't end with the answer, it starts with the answer. So I think the same thing, you run that model, you get that data point, you get that piece of advice and then just as we really discussed in the humans AI interaction piece, that's where the discussion starts. The model says this, how do we interpret that in the real world? How do we bring that to different stakeholders? Do you think your psychology background plays a role in your ability to communicate with people on this level? Yeah, I do, I do. I think spending four years thinking and studying that topic, it really helped me kind of frame up. It's always, for me, people first, mission always. So I always focus on the human element and I believe that in supply chain, more often than not, you're dealing with a lot of humans and those humans are oftentimes making decisions many times or augmented by AI that's running in the background but also getting them to understand your point of view is really about empathy, understanding their point of view and where they're coming from. That's absolutely the most important piece and candidly, I think a lot of times we hear this and they're entire businesses that are built on behavioral coaching. HBR is built on a massive empire on putting out publications that are based on behavioral strategies. That exists for a reason and so I think it would be foolish for any of us to ignore that that's an important thing. I definitely spent a lot of time discounting my psychology background early in my career. I wondered what the validity of that was long-term but I can tell you that I think it's just helped me have more meaningful decisions and more meaningful conversations as a result of that. In the teams that you manage and you work with customers, do you find yourself sort of between the sort of maybe more pure play data analyst and the sort of non-quantitative person? Do you find yourself working across those camps? Yeah, I do but I think one of the things that we focused on at Uber Freight and especially within the team that I'm fortunate to work with is we've really focused on building athletes, right? So folks that are skilled in data analysis and quantitative techniques but are also fundamentally sound when it comes to just having a good conversation with a customer or having a good internal dialogue and framing up a problem. I always hire for character first but also what I'm looking for is a passion and drive and malleability that they can come in and learn a variety of different things and bring a skill set and expand upon that. And so oftentimes when I'm speaking with a customer or having an internal conversation, the same person that is initiating that conversation from our side has also oftentimes written the SQL query, has run the analysis, has maybe done the forecasting and the modeling and they're oftentimes presenting their solution and the really fun thing about that is they get to own that in its entirety. And so I think kind of going back to your point, I think the quantitative and the behavioral points all have equal weight. And also when I think about it, the future of supply chain isn't just going to be one or the other, it's gonna be everything. And I think the folks who will do really well and will find like great deep satisfaction in their careers will have spent a lot of time developing techniques on one side and on the other. And I think they'll find that they're able to move the needle much more effectively than maybe somebody who has just spent a lot of time and over-indexed on one or the other. Absolutely, I kind of want to highlight a couple of things that you've said now. If nothing else, learners, just to help you feel a little better about all the time you just spent on your midterm, bringing up the SQL queries, you brought it up a few times. We know it's a pain to learn, but it has value. You're working with this kind of data set. You can gain a lot of experience just in the course so great to hear that. And I think what you just said about supply chain, being a business where you have to work with people is what draws, in my experience, all of us to it. Maybe some of us could have gone into pure mathematics or to other types of things, but solving problems, working with people is what makes it fun, really. Do you have any advice kind of where I was going to that question was advice for people that see themselves more strongly on the quantitative side. That's not all our learners. We have a number of hundreds, obviously, of learners, but I think maybe many of our learners already feel comfortable with the math or running the computer programs. What advice do you have for someone like that who wants to become an athlete like you described? Well, yeah, first of all, I would say that myself and a variety of other people envy you. Those are skills that I think you should be very proud of, first of all. The second thing is that you don't become an athlete in an echo chamber, right? You tend to have to, you have to just set some goals. You have to oftentimes step outside of your comfort zones and that sounds like a lot of platitudes, but oftentimes what that really comes down to is you can leverage your skills and you can come up with, you can identify a problem in your workplace or maybe even in your own personal life. Set the model, run the math, get to a solution and then just pitch it to maybe use somebody who sits next to you or spend some time with your manager, grab a one-on-one, have it, talk about it over lunch and really start flexing those muscles that you really wanna develop as an athlete and ultimately at the end of the day, it's really all about that first step and I think what you'll find is over time you'll get into a pace and a rhythm and you'll start learning to, I say this a lot, you have to learn to read the room and so you have to understand for the person that's sitting across the table for you or sitting around the room, they're gonna have a different perspective but you're never gonna know that if you don't ask and oftentimes think about your output as the prompt. So like right now we're seeing some questions come in. Ultimately at the end of the day, leverage your skills to create a good prompt and start that conversation. Oftentimes for the team that I work with, they'll create some wonderful analyses and then they'll say, what do you think? How about I walk you through this? Let me tell you where my head's at with this and I'll tell you those lead to some really good conversations and it's a great starting point. None of this is gonna happen overnight and it doesn't have to be perfect. I think you have to just accept the fact that you're growing as an individual and that's why all of you are in the program right now and if you didn't wanna grow, you probably wouldn't be here. So really at the end of the day, you have to just take that next step and leverage your skills already and just kick off conversation. Yeah, that is great advice, thank you. And it makes me think, learners in week seven, I'm putting some finishing touches on it now but you have what we're calling for the first time the field trip followup where I'll be giving you a really rather large database and asking you to apply some of the techniques that you saw on the field trip. And I think that's a great idea. If you come up with something, you see something in there, share it in the forums and let's see what your analysis says and let's walk through thinking about it just like you described, that's great, thank you. Believe it or not, we're starting to wind up and we have so many good questions. Maybe we could pull some off here. So thank you, Param for the question about blockchain internet of things. I think we got there. Chris has another one. I think you spoke to this a little bit but let's dive right in. The granularity level of forecasting when you're talking about forecasting, perhaps that next load of the driver. How zoomed in can you get, I guess, when you're coming in with the forecast? Yeah, well, so is Chris a TA? Yes. You're great, thank you. Thank you for all that you're doing. Thank you, Chris. So Chris, I want to point out this, when we think about initializing a forecast, oftentimes the dataset that you are offered either when you're new and you're new to the industry, you're oftentimes like a customer's dataset to initialize your forecast. So if a customer offers us a variety of things that we need to evaluate, we're gonna use that to initialize but oftentimes it's lacking some information. At best, you're gonna zoom in on maybe a weekly number of transactions that you are likely to move and that's at best. More often than not, you're forecasting just on a 365 day span when you think that you may have order of magnitude however many transactions you're gonna move in that year. It would be really exciting if when we think about like smoothing and exponential smoothing and things like that, it'd be really cool if we could get down to the level of appointment times that are historically found based on whatever facility and then destination receiving time. I think that ultimately at the end of the day would actually help us clean up a lot of the data because what you'll find is and you kind of switched this earlier that the amount of hours that a truck driver can drive today in North America is very different than what it was, I don't know a year ago, right? And so what we'll find is that those appointment times or that ability to forecast that that specific transaction will happen, it doesn't exist anymore. Like we wouldn't be able to execute that in today's world based on the truck driver. So the planning and scheduling, we typically rely on whatever the customer is informing us that their demand will be over the course of the year and the further zoomed in we can get, great. You would have some questions probably about the accuracy of that Chris. Like somebody tells you it's always gonna be at 11 a.m. on a Monday morning. I would wonder like, I would always push back and say what's your confidence interval on that? So yeah, so hopefully that answers your question. Yeah, thank you, that was really good. And I think really where Harris and I started talking a few months ago was thinking about getting the most out of that driver resource and two year point precisely, appointment times, moving them through as fast as possible. And now that that data has become digital, and that's where our conversation started months ago, there's some really exciting new opportunities there and through research and through just the community of CTO, we can keep those conversations going. I'm gonna jump down. I'm just gonna jump to John here. Integration options, is there an API that you provide to help people sort of plug into Uber freight data? Right, right. So we offer a couple of different channels. We've built our own kind of self-service platform. So we wanna give everybody the opportunity to plug in, whether they're large or small and they're able to plug in no matter what size they are. Whether they move one transaction a week or maybe one transaction a month, they're able to get access to the Uber freight network much in the same way that the enterprise customer would. The industry is built on the backbones of electronic data interchange. EDI, I hope many of you have heard about that. When you wanna talk about sticky technology, EDI is absolutely one of the stickiest technologies I've ever seen. It is everywhere. The real fun thing about that is that it gives you kind of like a point of reference. And so every time that we think about building a new API or like expanding our API, we just want it to be everything better than what EDI is. So API allows us to do a variety of things on your specific question that it was around, you know, tracking and billing and those things. Yeah, absolutely. Like that's kind of table stakes. The fun stuff really comes into play when you're actually starting to be able to leverage somebody of your own forecasting for procurement. And so if David comes to me and he says, hey, I would really like to go, San Antonio to Houston today. What's that gonna cost on the next Monday at 4 p.m.? Well, it's cool because I can like, I can run that in the system and then I can tell David and that's cool, right? But putting it in David's TMS system where he's already living natively and he's making a ton of decisions and evaluating a lot of things, leveraging our API to be able to plug that into your TMS. That's some of the cool stuff that we're doing now. And so based on that, you know, we're really excited about our ability to be able to kind of link up those things together and give people real, natural decision-making ability where they currently live. That is excellent. Thank you. And I promise I did not, I didn't either pay nor prompt Harris to say that. In week seven, we have content on EDI and TMS. So if that's an alphabet soup to you at this point, after one week, you'll be up to speed. Great question. Thank you. Thank you for that answer. Let's see. You know, one question, I wanna make sure we get to Azhar. Thank you so much for offering us questions before you even started. And one we were looking at, what happens if a driver steals a shipment? Yeah, man. Yeah, we obviously hope that that never happens, but I think one of the cool things about Uber Freight and our technology is understanding that things do happen and having good visibility for the life cycle of the shipment is super helpful. So when you think about a customer, like eventually finding out that their shipment has been removed from the network and is maybe in a place of disrepair, the cool thing about our network is that we have history on where that load has been previously. And then obviously we go through the normal channels of reporting that and making sure that the authorities are involved. So since we're not doing a lot of this to find out where our loads are, it's all kind of built in. The good thing about that is that there, I think customers have a greater sense of confidence with our ability to provide updates on where their shipments are and we're not relying on a third-party dataset or a third-party integration that is pinging and we're hoping that that works out. With our shipments, everything's real and it's live. And it's actually right there in the palm of hand because it's directly linked to that driver and that phone. Oh, interesting. Let's go to this question from Gurukant. How do you ensure that your driver partners remain profitable along with you? Yeah, I really like this question because it's actually at the core when we built Uber Freight. We launched in 2017 and we could've gone one of two ways. We could've said, look, we're gonna focus specifically on shippers and we're gonna try to solve all of their problems and we're gonna just focus on that. And then we'll eventually figure out the supply side. We led driver first, not only because we think that that is the right thing to do, but really because we recognize that carriers and especially owner operators, individual truck drivers are oftentimes disadvantaged from having access to a wider network because there are just multiple layers that they would have to work through to be able to have access to good paying freight. And so when we think about our driver partners remaining profitable with us, we're doing things like Powerloop. We're trying to create, trying to remove those barriers to entry so that they can, so that their earnings, more of their earnings stay alongside them and increases they leverage our platform more. I think more often than not, the fact that we're very transparent with rates and it's all visually, when a driver opens our app, they can see what they're going to be paid for a shipment. There's no negotiating. And so the good thing about that is that they know those things up front. There are no, there's no behind the scenes phone calls. There's no shaving of margin that we don't, that's not the purpose of this. We want to be fully transparent. We want our driver base to continue to grow along with us. And the way that you do that isn't by taking from them. It's about giving them more opportunity to earn more. Perom asked just following along on that, is Uber, how does Uber Eats fit into the Uber ecosystem? Yeah, yeah, no, so Uber globally really has multiple lines of business. The first one is obviously Uber rideshare, which many folks know and love. Then there's Uber Eats, which is a kind of its second standalone business unit. And that is really focused on aligning a courier supply with food demand, right? And making those actual transactions happen. Uber Eats is a really interesting function because it's actually a three-sided marketplace. So not only are you solving for the courier, the driver supply, you're thinking about the restaurants and making sure the restaurants are online. And they have the technology that they can handle transactions as well. And then you're obviously solving for the demand side as well. And so Uber Eats is kind of, it's a really interesting part of our business. It is interesting. And then obviously Uber Freight is a third standalone business unit within the Uber umbrella. Gosh. Guys, we are running up on our hour and you have so many great questions. Thank you. Are there any in here that you want to speak to? Make sure you get the chance to speak to before we go. So I think there's one down here. What kind of data do you collect to use in the AI model? So, yeah, no, this is, it's a great question. Just normally in the course of doing business within transportation and logistics, I'll toss out a couple of things here, but I think one of the things that really informs a lot of our AI and decision-making when it comes like surfacing options for a driver to select going forward. It's really origin, destination, time of arrival, time of expected unload. And those four points are really important for setting up kind of that decision framework simply is just really simply because you kind of have to have a buffer. So if they come in and they're a power loop load, right? This is just a drop in hook trailer that they pull in and drop off and they can go about their day. Their ability to transact their next engagement much faster, 30 minutes. If they're a live shipment where they're tied to that trailer and they back into a dock and they have to sit there for a while, their ability to take their next transaction or to engage in the next transaction is much longer, right? It could be variance of 48 hours just depending on how the facility is running. And so we're leveraging AI to really think about, okay, so the velocity of a power loop unit should be much higher. And so when we make a prediction on what those available loads could be, we really should be showing a different subset of loads that within kind of an hour or two of delivery, you should be able to go execute on. For a live shipment, you need to have a larger buffer in there. And so AI in the background is making those decisions and kind of evaluating that bit of information altogether. And obviously, you know, I mean, there are really some future constructs in this, right? When you think about like the weight of a load, for example, you're gonna think about what that gas and like the rate per mile really should be. So if I have a heavier load, my cost of gas is actually going to increase because I'm gonna be hauling it over a distance and the miles per gallon I'm able to achieve are probably much lower just depending on that weight. And so, you know, at some future state, should we have some consideration for weight of a load relative to the cost of gas and thinking about how we can maximize driver earnings from there as well? Just a thought. And I probably would have had that thought without that question. So great, great one. Yeah, thank you, Azhar. Well, we're running out on time. Thank you all so much for joining in. If we didn't get to your question, feel free to send it to me to put it to the forums. And as long as we can, we'll connect it with Harris and get some thoughts. I wanna do a little bit of housekeeping and it's bad news too. So we'll come back for some good news after I do this. But don't forget that your week eight graded assignment. So week eight, not right now, but week eight is proctored. Proctored means that you will download some software. You'll get very explicit instructions for this. And it will essentially ensure that you are following all of the rules and doing all of the work yourself when you do the week eight graded assignment. Completely upfront, I know it's a pain. I apologize in advance. The reason that we do it is that the comprehensive final exam, the test that you take to earn your MicroMasters credential is also proctored and we want to provide you this opportunity to practice with it, to make any mistakes you're gonna make with it now when it doesn't count for as much as it will in the CFX when you have only one chance to pass the test. So please be on the lookout for that. We'll give you plenty of information and as much tech support as we possibly can on that, but please keep that in mind. With that, I wanna thank Harris so much. Do you have any closing thoughts for our learners, people who are passionate about supply chain, who know that you work in an innovative company? Where's the future? Where should they be headed? What advice do you have? Yeah, I think that everybody will be surprised at what supply chain will look like even six months from now. I spoke of a company right here in Cambridge that's a startup. There are tons of opportunity to become involved and everything, your involvement doesn't always have to be on the scale of Uber and Uber freight, right? Your engagement could simply be working for an hour or doing some other things and leveraging your skills and your knowledge of the technology and pushing and asking those questions internally to really innovate where you're at. Not everybody has to make a leap and a giant leap of faith to a startup to make it all happen and see all the technology. There are a lot of things and hopefully many of you have seen, you're going through the MicroMasters that you can access and bring in the bear right now, today within your own working careers. For those of you who are thinking about a change of scenery, have some really thorough conversations outside of your organization because oftentimes what you'll find is that people in your lives, they're biased. So if you're talking to somebody in your family, they're gonna have a point of view. If you talk to people at your organization, they're gonna have a point of view. I would always encourage people to reach outside of your bubble to have some really informed conversations and I think you'll find that that time is well spent as long as you are coming into that with a good knowledge of kind of where your head's at. Thank you so much for your time, it's been really valuable. Thanks, David, really appreciate it. Thank you for tuning in and we'll see you at the next live event at the close of the course. Congratulations on completion of your midterm and we'll see you in the forums. Yeah, good luck.