Resources

Company resources

• Everstream Analytics company profile
• Everstream Analytics optimizing cold chain transportation
• Morgan Stanley announcement of Series B funding: Morgan Stanley Investment Management’s 1GT Co-Leads $50 Million Funding for Everstream Analytics
• Intergovernmental Panel on Climate Change (IPCC) Future Global Climate:
• Scenario-based Projections and Near-term Information chapter

Career resources

• KDNuggets (news and information site for Data Science, Machine Learning, AI and Analytics)
• Let’s Talk Supply Chain podcast
• Gartner Supply Chain podcast

Guests

Climate Rising Host: Professor Mike Toffel, Faculty Chair, Business & Environment Initiative

Jim Hayden, Chief Data Scientist, Everstream Analytics

Transcript

Editor’s Note: The following was prepared by a machine algorithm, and may not perfectly reflect the audio file of the interview.

Mike Toffel:
This is Climate Rising, a podcast from Harvard Business School, and I’m your host, Mike Toffel, a professor here at HBS.

In today’s episode, I’m talking with Jim Hayden, Chief Data Scientist at Everstream Analytics, which is a data science company that helps its clients manage supply chain risks, including those from climate change. His company is using artificial intelligence and machine learning to help predict supply chain disruptions from extreme weather as well as provide longer-term scenario planning.

I’ll ask Jim about how Everstream helps companies adapt and respond to climate change risks, and why AI is an important tool for optimizing supply chains. And, as usual, I’ll ask him to share some advice for those interested in working at the intersection of business and climate change.

Here’s my interview with Jim Hayden from Everstream Analytics.

Jim, thanks so much for joining us here on Climate Rising.

Jim Hayden:
Nice to be here, Mike. Looking forward to the conversation.

Mike Toffel:
Terrific. Well, let's start with an introduction. What's your role at Everstream?

Jim Hayden:
I'm the chief data scientist here at Everstream Analytics, and in that role, I manage a few different teams of data scientists working in different problem areas. I manage our applied meteorology team. I manage our intelligence solutions team, which are human beings doing research into global incidents. And I also manage our product team.

Mike Toffel:
So Jim, tell me a little bit about how large your company is and where it's based?

Jim Hayden:
Sure. We're a completely remote company. We've been that way since pre-COVID. I have people in Australia and Singapore and India and Germany and Alaska and the US. We've got a couple hubs, where there're people, based on where the prior acquisitions came from. In Germany, we have a good team there, and in the US in the Florida area, we've got a pretty good size team there. Other than that, we're all over the globe.

Mike Toffel:
Wow. That sounds like a pretty challenging situation to manage many teams around the world. What are the biggest challenges for you and what have you found to be successful?

Jim Hayden:
Sure. It is a challenge, and the social connection is a big one. Right? And so we do things like have happy hours once a month, but they're remote happy hours. We do things like try to get people together in small teams, at least the people they work with, fairly regularly and that helps as well. And what we found is those employees that have less experience in the office, that maybe are new entrants into their career, they seem to have a little harder time. There's none of the, well, how does work work that they're learning, right? And had you spent time in the office, you learn some things about work that aren't written down anywhere.

Mike Toffel:
Yeah, the hallway conversations that you don't plan a meeting for, but you just see people around.

Jim Hayden:
Exactly.

Mike Toffel:
Yeah. And how did you get there? How did you land at Everstream?

Jim Hayden:
So fortunately, early in my career, I figured out how important data was. And so I started out as a database administrator, then a database architect, and then I had some management skills. And I got put in charge of a team that had PhDs in something called machine learning that I hadn't heard about before. And this was in the mid-90s. And our task was to build the surveillance system for the NASDAQ stock market. So I jumped right in and got an introduction to how machine learning works, how it works on big data. And I've been doing that ever since. Have done it throughout the financial services industry on different problems like anti-money laundering. I've done it in telecommunications sector to optimize networks. And now for the past nine years I've been applying various machine learning techniques to help supply chains get optimized.

Mike Toffel:
And so what led you to think that supply chains was the next stop for you in your career?

Jim Hayden:
Well, I've always been chasing bigger data. And when I understood that people were using IoT, internet of things, devices on cargo containers to track them around the globe, I figured that's a lot of data, and I bet we can do some interesting analytics on top of that.

Mike Toffel:
Terrific. So what's the elevator pitch for what Everstream Analytics does?

Jim Hayden:
Well, we use machine learning to help our customers understand their complex multi-tier supply chains and help them mitigate risk and disruption. We also help them improve their ESG performance, but the essence of what we do is we tell our customers something they don't know that they should know about. And that can be along many different dimensions using forecast, using predictive modeling, using human intelligence, just looking at newspapers that are local in foreign languages to really let them know what's going on and could impact them with as much notice as possible.

Mike Toffel:
And so what types of decisions are you trying to influence for these companies?

Jim Hayden:
First, it's to let them know that there's risk ahead of them. Then it's to help them understand their options to mitigate the risk. And in certain areas, we're actually helping decide, using AI to decide, what they should do about the risk.

Mike Toffel:
Got it. And what data do you use to help build these models that are these predictive models that help expose these risks and help figure out what the options are and then perhaps even recommend one of the options for them to choose? Where are you getting this data?

Jim Hayden:
Well, no matter how you look at it, we're a big data company. We get data from all over. On our forecast data alone, we get over a billion data points a day. We capture that from the National Weather Service, the European Weather Service, from different platforms. And our meteorologists take those models, they take them apart and put them back together again to get a little more specific on supply chain risk. And that's just one of our sources. We look at a million news and media posts in an hour to understand who, what, and where could be impacting the world's supply chains. We capture shipment data from our customers with tens of millions of shipments on who's trading to whom around the globe, what modes of transport they're using. And that type of information allows us to get to understanding things like carbon emissions associated with shipments or outbound goods delivery.

Mike Toffel:
So you're gathering information on weather forecasts, news and media reports about current and future events, and you're sort of putting on top of that the transit that your companies are engaged with with their shipments and where things are coming from, where they're going to, which I imagine also overlays with sort of what are the products and services that they are requiring and that they're providing and putting all that together into various models. Let's talk a little bit about the modeling. So you have all this data. If you think about your old equation of these are all the Xs and then you're trying to predict a few Ys, which is the dependent variable in an equation, what are you trying to predict in these instances?

Jim Hayden:
Like I mentioned, I have a few different data science teams. So one of them is focused on predicting arrival times of shipments based on prevailing conditions. They take the proposed route that the cargo is going to take, and that can be fairly complicated. It can be truck to rail to ocean to rail to truck to finally get there. And we overlay a weather forecast across the entire route. And that allows us to understand potential hazards as well as allows them to help optimize the equipment that they're using. So not only do we look at timeliness, we can look at the quality of the delivery. Certain goods need to be temperature-controlled. We can look at the carbon emissions associated with it and help our customers optimize on that. And of course, we can look at cost if our customers want to give us that data.

Mike Toffel:
Okay. So that is predicting arrival time, quality, carbon emissions, cost. Okay.

Jim Hayden:
Right.

Mike Toffel:
So that's all overlaying the weather forecasts with the shipment, knowing the shipment modes and the locations.

Jim Hayden:
That's right. Right. Another important thing we do for our customers is we give them visibility into their sub-tier suppliers. This is a big problem in supply chain.

Mike Toffel:
Yeah. So let's talk about that. So sub-tier means beyond the suppliers from whom they directly procure. Is that right?

Jim Hayden:
That's right.

Mike Toffel:
So tier-two, tier-three, tier-four, in other words.

Jim Hayden:
Sure. And these are blind spots. And so another big data source we have is we have over 100 billion import/export records that we've gathered from third parties and open sources, and that tells you who's shipping what to whom. And we take that information, and it's very dirty data. We use machine learning to help clean that data up. As you can imagine, these come from custom forms, where people are typing in the names of companies. They're in foreign languages, they're abbreviated, and we apply what's called entity resolution to that. And that's a big topic in AI today, figuring out who's who and where exactly are they trading these goods. Once we have that, we build out what we call our knowledge graph. We can then take our customer's tier-one suppliers that they know and start understanding who's trading with them and then who's trading with them, who's trading with them to get to the true flow of goods in the value chain all the way down to the raw material.

Mike Toffel:
Now, these records of trades, are these US-only records? Are they import/export-only records? Do they also cover domestic? What's the limitation and contours of that?

Jim Hayden:
Yeah. We take all sorts of data to fill this in. Primarily, it's import/export records, and that's across about 160 countries.

Mike Toffel:
Wow.

Jim Hayden:
And that tells us who's trading with who, certainly all the major countries, but then that still leaves you, like you said, intra-USA shipments and trades and intra-EU shipments and trades that don't cross a border. And so we use shipment records to figure that out from our customers. We're also using a few other techniques to understand patterns in trading relationships. Maybe it's truck movements along certain corridors that go back and forth, and that can show us a trading relationship there. We're constantly in the hunt for new data to be able to determine these relationships and fill in all of those gaps.

Mike Toffel:
Interesting. So what are the other types of decisions? You mentioned arrival time and quality and carbon emissions and cost. What are other types of decisions you're trying to help customers get a handle on?

Jim Hayden:
Well, certainly, the impact of climate and whether that's a two-week very granular forecast or whether that's a 50- or 100-year estimate on what the future climate looks like for certain locations. And that's really important for our customers as they look to deal with things like heat stress and drought stress and precipitation stress in the future, sea level rise, tropical cyclones and wind speeds. And our projections are based on different impact scenarios, depending frankly on how human beings behave over the next 10 or 20 or 30 years to help with climate change, and we have severe to optimistic, let's call it. And that type of information, you can make important decisions about where to put your assets, where to do your trading, where to have your suppliers. And people often underestimate heat stress and how important that is. You can't work in warehouses when it's 130 degrees out. These are important decisions, more planning-oriented than they are actual execution and operational.

Mike Toffel:
So you're helping companies think about siting decisions?

Jim Hayden:
Exactly.

Mike Toffel:
Like where should they put their operations warehouses and so on. And it sounds like also sourcing questions.

Jim Hayden:
Right. Where they should procure from. And that is not only tier-ones. If they find out that several of their tier-twos or tier-threes are in high-risk locations, they should start looking at moving or alternate sourcing decisions related to that too.

Mike Toffel:
Now, you mention there is a range from severe to optimistic scenarios. Right? So I think sometimes people refer to this as scenario planning or scenario analysis, where you have different scenarios, where in some cases you imagine, okay, lots of mitigation's going to occur, the best case scenario, and then you have sort of a worst case scenario, business as usual, we're going to go back to coal and so on.

And then you have a variety of intermediate scenarios, and the IPCC reports put out by the UN articulate a variety of these scenarios. I imagine you can't give them too many scenarios because it'll make your head explode. So imagine you choose a subset of these many scenarios that are possible. How then do you help your clients think through the trade-offs between these scenarios? Because I imagine under some scenarios, it will say site here and source there. And under a different scenario, very different recommendations. How do they think about which scenarios to follow and how to balance this?

Jim Hayden:
Yeah. Some of that depends on how sophisticated the customer is. Some of them are actually looking 100 years out, whereas a lot of them, their view of the world is five, no more than seven, years out. We try to keep it simple. We try to keep things red, yellow, green on risk and then take it from there. So if you just show somebody red under any of the three scenarios, but from moderate to severe and it's red, that's the first thing you should worry about. Right? And in this world of supply chain risk, that's kind of what we're focused on too. It's helping them prioritize what they should worry about. And then that's the purpose of something as simple as red, yellow, and green.

Mike Toffel:
Yeah. So say a little more about some companies are thinking five to seven years and some are thinking 100 years out. What are you noticing about the different types of companies or management teams or industries that differentiate their time horizon in thinking through these types of issues?

Jim Hayden:
Yeah. Certain industries we see just move a lot quicker. The ones you'd expect, like the biopharm industries, they're quick in everything they do. So they're thinking a little more near term, a little more 10 years out max. Whereas some of the more established and blue chip companies, like in the chemical industry, they're thinking longer range, from what I've seen.

Mike Toffel:
And is this partly because of the magnitude of the investments that they have to make?

Jim Hayden:
Yeah.

Mike Toffel:
If you're making a $1 billion investment in a chemical plant versus a $10 million warehouse or something like that, does that also play a role?

Jim Hayden:
Yeah. Absolutely. And that's exactly it. And where you should have hazardous material or not, that comes into play. Where you can transport hazard materials and where you can't do that comes into play for that industry.

Mike Toffel:
I see. So let's talk through a couple of examples. Now, I know you've done some work for AB InBev in Texas. I wonder if you can talk through what were the types of risks that you helped them identify and what decisions did that lead for them?

Jim Hayden:
Sure. So we help them manage ongoing risk to their goods being transported. I mentioned this a bit earlier, where we'll understand the route of their shipment, we'll overlay the weather forecast, and then we do prescriptive analytics here. Not only are we predicting what the ambient temperature will be, we're prescribing the optimal equipment to use. And in this case, we're telling them whether they need temperature-controlled equipment or not. And what companies used to do is say, okay, for the hot months, from June through August, I'm always going to use temperature-controlled equipment. Well, you don't need to do that. And every time you don't do that, not only do you save 50 cents a mile, but you save significant carbon emissions as well. And so that's a fairly straightforward one that's integrated into their operations and works pretty seamlessly.

Now, a more impactful one that we helped them with was with the polar vortex in February of 2021, and this deep freeze in Texas. And our applied meteorology team identified some stratospheric warming event in January of 2021. They weren't exactly sure what was going to happen, but they know it destabilized the polar vortex. And then three weeks in advance, they got a little more fidelity on the model, and they understood it was going to get cold in Texas, still giving our customers a warning, but it wasn't red yet. And then two weeks in advance, we understood this was going to be freezing, so it's red. We still didn't understand how long the duration was going to be. And then a week in advance, we told them that there was going to be a significant freeze event here, could go all the way down to Mexico. And that allowed them to take action.

The action they took was to halt pre-loading and positioning of trailers to the brewery grounds one week in advance. No beer in the lot in Texas at all after shipments. And they accelerated some shipments outbound to wholesalers. Any trailers that weren't able to be picked up, they backed them up next to the brewery, opened the loading dock doors so they could get some of the hot air out of the brewery that it was giving off. And that allowed to save them, well, over $1 million dollars, let's put it that way.

Mike Toffel:
This is because you're worried about frozen inventory, essentially destroying the beer.

Jim Hayden:
Right. A deep freeze to something like beer can be pretty significant.

Mike Toffel:
Right.

Jim Hayden:
That was the first time they had done anything like that in our 10 years working with AB InBev to just use data, believe a forecast, and take action ahead of time. And that saved them a lot. And the impact of that polar vortex is still being felt a little bit today. People understood that oil and gas production was halted. What not everyone understood is most of the plastic in the world uses some type of petroleum, and most things in the world that are in a manufacturing process have some plastic in it. And so that impact was pretty significant. And we had a couple customers that were able to notify in advance of this, actually a bottling company, and they were able to source elsewhere.

Mike Toffel:
You're saying this plastics issue is still being felt because they didn't take evasive actions early enough.

Jim Hayden:
Exactly. Yeah. And they're still feeling the impact of this, of not having that. When you lose inventory coming out of an area for four to eight weeks, the ripple effect on that can last months and months and months too.

Mike Toffel:
Now, how sure was your team in the advice that they gave AB InBev? Because if the polar vortex had not transpired in the way that you had predicted, presumably all the actions that they took would've been costly and they'd be looking at you saying, what? What happened?

Jim Hayden:
Right.

Mike Toffel:
You said this was going to happen. It didn't happen. If you make those misfires a few times, then I think the credibility may unravel. But on the other hand, you don't want to wait till you're 100% sure because you're never going to be 100% sure. It'll probably be too late for them to take evasive action before you're 100% sure. So how do you balance this sort of type-one error, type-two error is one way to characterize it?

Jim Hayden:
Yeah, yeah. No, that's a great question. How do you balance the false positives versus false negatives and what's the right percentage there? Some customers might be okay with a few false positives, but I don't want to miss anything. So some of that depends on the customer, but what we really try to do is be as transparent as possible. So in our world, risk equals probability times severity. And so we knew it was severe, that the probability of it hitting exactly there wasn't 100%, but as it got closer to 100%, they saw the potential risk score going up and up and up. And in our world, that's an indicator that it's going to happen.

Mike Toffel:
Got it. So you're transparent with your clients so that they can apply their own comfort of risk tolerance.

Jim Hayden:
Right. Exactly.

Mike Toffel:
Right.

Jim Hayden:
Yeah, that's key in everything we show our customers. We try to give them as much context as possible, the source of information, what we believe the impact will be, so they can again make their own decision. And some of them have their own playbooks, depending on the severity, and the probability of this event.

Mike Toffel:
Got it. Okay. Great. So that's the AB InBev story of predicting this weather event and helping them prepare for it by, for example, draining their inventory out of their lots, either by sending them off or by backing their trucks up to the brewery to sort of help heat them. What are some other examples that you have thought through? I know with the medical device industry, you've helped them become more resilient to wildfires by helping them identify risk, not only to their own establishments, but to their suppliers and their supplier's suppliers. Can you tell us a little bit about that?

Jim Hayden:
Right. Yeah. That's using a combination of data sources. We derive these sub-tier relationships to be able to show our customers their raw material all the way to manufacturing value chain. And once we can do that, we can look for risk in that sub-tier. And so in this case, more than one of our medical device customers has used us to do their sub-tier mapping. And the volume of suppliers you then need to monitor becomes fairly big fairly quickly. I'll give you an example. So they had 29 tier-one suppliers just for a handful of their products. That turns into 212 tier-two suppliers. That's over 1700 tier-three suppliers. That's then 14,000 tier-four suppliers. And you can imagine the number of locations we need to start monitoring for that. So we're a big data company. We have that ability.

When we see something like a wildfire in the west and it's growing and the winds are picking up, we can see what's probably in the path of that. And we're able to alert both of these customers that there was a tier-three supplier, two tier-three suppliers that were in the direct path of this wildfire. What that allowed them to do, because there's a delay in every tier you go down, by about a week's worth of inventory on hand, that allowed them to understand what the impact was going to be and move forward some of their orders from the tier-one that were downstream from those tier-three suppliers.

Now, the tier-one didn't know this risk existed yet either. So as far as they're concerned, they're just getting move forward orders from a customer and that's good, that's more revenue, but it's an indicator of how the first to know can often take this mitigation strategy and win. So it's not only tell me what the risk is, but tell me as soon as it's possible to know about this risk so I can be one of the first to mitigate. Because if everyone knew about it, they would all try to move their orders forward. They wouldn't have been able to do that.

Mike Toffel:
Now, medical devices, what are examples of tier-three suppliers to a medical device company? So medical devices are often made of plastics and other synthetics. Just take us through a sample supply chain just to put a little bit more clarity on what we're even talking about here.

Jim Hayden:
Yeah. Sure. So our customers often tell us what the end product is. And in this case, it was artificial joints, so an artificial knee and artificial hip. And these were ball bearings that were used to simulate the rotation of the joints. And that was what was in the line of fire for them. So a component in the tier-three that goes into a component in the tier-two and then the tier-one gives them the final product, they were not going to be able to have one of the key components, which is, I'm not sure what it's called in the medical device world, but if you look at a hip, you know what I'm talking about.

Mike Toffel:
Got it. Interesting. Okay. So essentially, for example, ball bearings are a tier-three of a medical device, which they worried was in the path. So in that case, they accelerated their tier-ones. They said, hey, hurry up, we actually want more volume.

Jim Hayden:
Right.

Mike Toffel:
I'm just going to call it a ball bearing. Before the ball bearing shortage hit their tier-twos and then, ultimately, their tier-ones, they're like, let's get a jumpstart on this.

Jim Hayden:
Exactly.

Mike Toffel:
And you could have also perhaps, or they could have also perhaps said, where else can we get ball bearings on a short order?

Jim Hayden:
Right. Yeah.

Mike Toffel:
We have a three- or four-week lead time now. Where else can we procure them in advance?

Jim Hayden:
Yeah. And sometimes they'll actually tell their tier-one supplier where they can get, based on our information, an alternative sourcing location.

Mike Toffel:
Super interesting. Okay. Great. And so that's helping a company reach into their, in this case, tier-three suppliers or supplier-supplier-supplier to assess the risk and then giving your client advanced notice of that. You've also done this with regard to water levels in rivers that lead to ports which affect manufacturers' outbound shipments as they're trying to bring their products to market. Can you tell us a little bit about that?

Jim Hayden:
Sure. Not only do we do long-term water level risk using climate models, but we do short-term monitoring as well and prediction as well. And in 2022, there were some record low water levels on key waterways, such as the Rhine River and the Mississippi River. And when they get too low, they're unable to load as much capacity on the vessels going up and down the river because it takes them deeper into the water. But there are certain vessels, barges, that have a much lower water line. And so here's an example again, where because we gave early notification to our customers, one of their mitigation approaches was to then use barges for their shipping. They were first to get the barges. The capacity of barges soon started to diminish, but the customers that were the first actors to get them had the advantage of being able to ship their goods on these special barges.

Mike Toffel:
Got it. So they got earlier access before there became this crunch on barge capacity.

Jim Hayden:
Exactly.

Mike Toffel:
They were able to secure it.

Jim Hayden:
Yep.

Mike Toffel:
What's the competition look like in this space? I mean, you mentioned a lot of the data sources that you're drawing from are really publicly available in the sense that there's national weather services, there's news and media, the import/export documents, a lot of those are public as well. What prevents other companies from scraping that same data sources and coming up with pretty much the same models? How do you differentiate from other companies who are in this space?

Jim Hayden:
Yeah. we get asked that all the time. And so one of the things that differentiates us is our end-to-end view. So a lot of our competitors are either just procurement, just source to make or just make to distribute, make to deliver. We do both so we can understand their entire value chain and understand the risk associated with it. But from a data source perspective, we feel having our own applied meteorology team gives us a leg up there. We have certain partnerships with third party logistics companies that have unique access to some shipment data that they're allowed to share with us under certain conditions. That helps then fill in a lot of the gaps that maybe our competitors might have. And I've got an innovative data science team that manufactures some of their own data, looking at news and media to understand what's going on and what might be happening in the future.

Mike Toffel:
Got it. So some proprietary access to data and then the rest is sort of modeling skill, an interpretive skill set.

Jim Hayden:
Yeah. I mean, you're right. We're using open source, public domain algorithms, machine learning algorithms, for most of this. What then makes your models better are the data you put into it and the data scientists trying to build those models. We feel we have some unique combinations.

Mike Toffel:
Yeah. So in this space, and I was talking to someone from another company that is doing flood predictions, and I was saying, well, isn't there going to be some pressure for that to be a government function? And so here, too, that question comes to mind. If you're giving better advice in advance, for example, of the freeze in Texas and those who followed your advice saved millions or in some cases it could be even a lot more than just a few million, if you're protecting the infrastructure of oil rigs, for example, or complicated manufacturing companies like chemicals, if the headlines after those storms are these companies saved themselves by subscribing to this data service and these companies had billions of losses, which is going to lead to unemployment and so on, I imagine at some point there might be some pressure to say, hey, why isn't the government stepping in to do this type of projection and hire a data science team and make this information more available?

Do you expect that that will be a response that you're going to see as disasters become more headlines and as the winners and losers of those disasters become more salient based on this data story?

Jim Hayden:
Yeah. I think we'll see some of that. I think we're already seeing governments step in on a lot of ESG-related regulations. And an example of that would be forced labor. And now there are regulations for looking not only at your tier-one suppliers for use of forced labor, but in your sub-tiers. And so that helps us. We had our client council meeting a couple weeks ago. And they talked about sharing this data. It's for the greater good of the industry that we find these bad actors. And as I mentioned earlier, they share some of the same tier-two, tier-three suppliers. So why wouldn't we help out? And this is something we're considering too, to just help out industries for the greater good, looking for this type of behavior. And I could see the government stepping in, doing same thing for climate risk and other types of ESG violation.

Mike Toffel:
Yeah. I mean, in a way it would be an extension, for example, of the idea that weather services are national. There are government investments in predicting the weather. Maybe there'll be more government investment in some of these areas that you're engaged in.

Jim Hayden:
Yeah. If it wasn't for the National Weather Service and the European Service, we wouldn't have the models we have today.

Mike Toffel:
Yeah. Interesting. So what's next for Everstream Analytics? Where do you see the next class of problems that you're going to try and help clients address or new geographies you're going to try and tackle or new industries? What are you seeing as the future?

Jim Hayden:
Well, we just announced a series B funding of 50 million, which was co-led by Morgan Stanley as part of its 1GT private equity platform. So that's to remove one gross ton of carbon emissions by 2050. So they saw that we were heading in that direction, optimizing logistic shipments based on carbon emissions. And it's not just optimizing on should I send this ocean versus air versus truck versus rail. It's understanding with a little more fidelity what the impact of these carbon emissions are and what's really producing them. So the transport of hazardous material produces much more carbon emissions than just a regular carbon container full of goods. The amount of movement of these containers at a port can determine how much carbon emissions. Congestion at a port, if a vessel has to anchor outside a port for two or three days. That's typically not considered in these carbon emissions.

So we're getting deeper and deeper into that to let our customers not only understand the impact of these emissions, the generation of these scope-three emissions, but to offer some mitigation strategies, to let them trade off, to let them maybe show their customers that they're trading off. I'm paying a little more for this shipment. Look what it's done to reduce carbon emissions. Right? Some of our shippers think that way and they want to promote this information.

Mike Toffel:
Right. I can imagine, for example, if you're shipping, if the weather conditions over the past few days or other labor strikes have created congestion in the port, your analytics could instruct the ship to slow sail rather than speed up and then anchor for five days, to slow sail, saving energy along the way. And then the products will be unloaded sort of at the same time anyway.

Jim Hayden:
Yep. That's an example. Or we could tell our customers that if the cargo hasn't been loaded on that ship to the West Coast yet, put it on the ship to the East Coast and use rail to get it to its final destination.

Mike Toffel:
I see.

Jim Hayden:
So those are the types of decisions we can help them with.

Mike Toffel:
Very interesting. Great. This is our favorite last question. Some of our listeners are considering dedicating their careers in business and climate change in some manner. What we've been talking about here has been largely technical, using data to build models. Some of our listeners are going to have those skills and some of them won't, but sort of think this is an interesting area. What do you see as opportunities for both folks with technical data science backgrounds and for those who don't have that orientation?

Jim Hayden:
Sure. Yeah. Well, it's certainly a great time to jump into AI. It's all over the place now, and it's moving faster than I've ever seen it. Like I said, I've been working with it for 25 years. And the speed with which new inventions are coming out, and these are inventions that can add value too, like ChatGPT. We've got data sciences experimenting with its potential value on some of our use cases. So don't be afraid of it. It's not Terminator. It won't be for a long time, not in your careers anyway. So jump in. Don't be afraid of AI. You don't really need to understand how the algorithms work to understand AI. Just understand the types of problems that it can solve.

But at Everstream, it takes a village to do what we're doing. We have not only applied meteorology teams, but we have research analysts, we have customer success teams that talk to the customers about potential litigation strategies. They don't need to understand how the AI works or how the forecasts work, but they do need to understand supply chain and then how they can help the customers mitigate the risk. So there's plenty of opportunity around combining data with combining an understanding of supply chain and working with customers and the industries to help mitigate their risk.

Mike Toffel:
And how do companies find opportunities at companies like Everstream?

Jim Hayden:
So we hire all sorts of people, actually certain types of people. We like curious people here at Everstream. We're solving lots of problems. Some problems that we're going to solve we don't even know about yet. And so one thing about data scientists is they understand the data science part of it, but they're generally pretty hungry to apply that to different problems over time. They don't want to be stuck in one problem. We don't have researchers on new algorithms in our company. We have applied machine learning scientists that solve problems with machine learning.

Mike Toffel:
So are there conferences or websites or listservs that folks who are interested in this space should subscribe to or attend?

Jim Hayden:
Sure. Sure. So from a website perspective, if you're more technical, look at KDnuggets. Before it was called data science, it was called knowledge discovery. So it's KDnuggets, and they've been providing insights into new algorithms. It goes from a little bit technical to really technical, so that's for that audience. And then there're plenty of great resources out there. I follow Let's Talk Supply Chain with Sarah Barnes-Humphrey. the Gartner Supply Chain Podcast is interesting to me as well. And yeah, other than that, don't be afraid of the technology. You don't need to understand the inner workings of it. Just understand the business problems it can be used to solve.

Mike Toffel:
Great. And we'll put links to all those materials in the show notes. Thanks for that laundry list of ideas.

Jim Hayden:
Sure.

Mike Toffel:
Well, Jim, it's been a terrific, really insightful interview. Thank you so much for sharing your time with us and sharing the story of Everstream Analytics with Climate Rising. Appreciate it.

Jim Hayden:
It's been my pleasure. Thanks for having me.

Mike Toffel:
That was my conversation with Jim Hayden, Chief Data Scientist of Everstream Analytics.