Jannine Krish:

Hello, everyone! We are just waiting for everybody to join. We're gonna give it a couple minutes, so please stand by.

Thank you for joining. We'll just give it one more minute as we wait for people to come in.

Seeing lots of… lots of participants from around the world.

Okay, let's get started.

Thank you, everyone, for joining us today, and welcome to today's webinar, Global Trade Intelligence in the Age of AI. I'm Jannine, Chief Marketing Officer at ImportGenius, and I'm going to be moderating today's session. We're thrilled to have you with us!

As we know, the world runs on supply chains, but too often, supply chain strategy is not being given the attention it deserves. And in today's environment, it's becoming ever more important, and really belongs

At the center of the boardroom.

With AI and everything around that, you know, it's not just about having the tools, but it's about knowing how to use them to give you that advantage. And not just reacting to disruptions, but actually finding those golden opportunities within your supply chain.

Our customers are asking us the same questions around AI. How can it actually give us this advantage?

So today, we're going to show you not just the theory, but actual practical ways that AI and some of our other features that we've dropped this quarter

are really changing the game for trade intelligence. From risk detection to supplier discovery, from better product insights to lead generation, there's a lot of different use cases that we can show you. Today, we'll focus on just a few.

But I'm gonna… we're gonna spend the next 45 minutes, giving you a presentation from James, our Chief Product and Technology Officer, where he's gonna walk through.

Ho we're building AI features that are very practical and responsible.

And then Dan is gonna give you an actual demo of how the, you know, how you can use these new tools and some of our new features in action.

And I'll just… I'll give them a moment to introduce themselves. But first, I just wanted to let you know there's also an opportunity to ask questions during this webinar, so at any point, just plop your questions in the Q&A at the bottom.

And we will get to them at the last 10 minutes of this webinar.

And if at any point you decide you would like a personal walkthrough of the tools we're presenting today, and how it can give you more visibility into your supply chain, you're welcome to book a 20-minute demo with us. We're going to be plugging, the link in the chat in just a moment. Okay.

Without further ado, let's give a moment to our speakers to introduce themselves. James, over to you.

James Orr:

Hi folks, James here. I'm the CPTO at ImportGenius, and I run the product engineering and AI teams that conceive and build our product, as well as managing our data, which is the core of our business.

Daniel Mikhail:

Awesome, thanks, James, and I'm Daniel Mikhail. I'm the Enterprise Sales Lead here at ImportGenius. I've been in supply chain for the better part of the last decade, from four different perspectives, so I started in compliance and sanctions, moved into critical minerals, then logistics, and now in trade data.

Looking forward to showing you all a demo of our platform here today.

Jannine Krish:

Thank you both. Now, before we jump right in, I do want to start today's webinar with a poll.

So if you could all take a moment…

To answer this question, what shipment-level use case would help you the most?

So if you could take a moment to answer this, I'm just gonna give it 30 more seconds, and I'm going to share the results.

Okay, lots of you are answering.

I can still see people answering, so let's just give it 10 more seconds.

Great, thank you very much. You can see quite a variety here. Many of you really want to monitor your competitor shipments and who they're buying and selling to.

Great, thank you for, for participating in that.

James, why don't I hand it off to you?

James Orr:

Sure, thanks, Janine.

Alright, welcome, folks.

Here at ImportGenius, we've brought out a series of webinars throughout 2025, and the audience keeps growing, with a lot of folks continuing to tune in, so to our regular viewers, welcome back. We're honored that you're choosing to spend your time with us.

In return, we're going to do our best to make it worth your while, keep you up to date on ImportGenius, but also the larger world of trade data technology.

Now, if you're new to this series, and you'd like a brief intro to the larger world of trade data, I recommend our spring webinar that Daniel and I also did with Janine. In that one, we go over some of the fundamentals.

In that spring webinar, we also launched our first integrated AI feature, the Genius Company Profiler. So this is our deep research tool that scrapes information from the web and combines it with our trade data to investigate companies, and that is one of the primary activities that our users do.

So, since we last spoke, the world of AI has continued to progress at a dizzying pace. OpenAI released the long-awaited GPT-5. At first, people didn't really like it, but now it's widely recognized as a major leap in capabilities, both on the consumer chatbot side, but also coding and engineering.

Meta, just this month, released a new category of device. These are smart glasses with a private display inside the lens and a wristband that uses the subtle movements of your hand to control the interface.

Will meta display glasses become the new way that people interact with AI? Will they take a bite out of the smartphone business?

Probably too early to tell, but Mark Suckerberg looks pretty cool on them.

We've got AI labs across China are continuing to dominate the open source world. So along with the now-familiar deep-sea models that shook the stock market in Q1, Alibaba's Coin Labs, Moonshot's KimiK2, and Z.ai GLM have been consistently holding and advancing the state of the art, sprinting way ahead of the West.

And this new tech is much of the focus of humanity's attention and efforts, not to mention our capital, worldwide. And the resulting pace of innovation on this frontier is dizzying.

And with the announcement of America's AI action plan from the White House, it's starting to look like a space race, you know, with competition drawn along national lines.

With the fuss over chip export restrictions, and now China's refusal to import any new goods from Nvidia, this whole story's starting to spill over into our world. And of course it is, because trade is behind everything.

Now, every tech company, ImportGenius included, is eager to figure out how we can use this new cutting-edge stuff.

But there's still plenty that AI is no good at. You know, I think we've all experienced the frustration of watching modern AI profoundly fail at simple tasks. You know, here the users ask for a map of Europe that shows the most dangerous animal for each country, and, you know, the model comes back and proudly presents to you something that makes no sense at all.

But there's…

So we can't blindly embrace this capable but imperfect technology. We have to look at it as one tool in the toolbox. You know, it's weird that one of the tools in the toolbox is, like, a digital simulation of human thought straight out of a sci-fi story, sticks out next to the hammer and the saw.

But we find that in many scenarios, there are places where this technology fits. It solves a problem, it provides unique power and value within fine boundaries.

We can't just set modern AI loops on our application, or on our data, because it can do things wildly wrong, and because we're a data company, trust is key to our business, and the way to create data products that people trust is to carefully control the processes that manage that data.

You know, that trust is why our research department's able to work with global media organizations and provide research and sources for journalists. You know, we're proud to have built these relationships with the media, and I think it speaks to how important it is these days to have reputable sources of data in a world of misinformation.

And we get contacted daily by journalists worldwide, looking for verifiable evidence that they can use as a source.

And similarly, we often get used in legal proceedings for the same reason.

And yes, we were on the ground floor of the radioactive shrimp story that broke this month. Our research department was able to provide key shipping records that supported that journalism by the Associated Press.

We've also continued to publish our own journalism, Breaking Stories and the Manifest, where you can get that delivered to your inbox if you sign up on our website.

So, how do we build trust in our data?

As an enterprise data company, we practice the discipline known as data governance. This is a big subject, but it ties into the presentation today, so I'm going to touch on it briefly.

Trade data is messy and imperfect. It comes from organizations all over the world, different languages, using different standards. It's a complex process to develop international relationships that allow us to acquire data close to the source and make sense of it, load it into our infrastructure, and ultimately serve it up to our customers in a usable format.

So when a user logs in to ImportGenius and looks at a row and a table, they're seeing information that's made a long journey.

Now, those rows on the right didn't necessarily start as a perfect digital record. It might have come to us as a data dump from the software used at a customs agency somewhere in Eastern Europe, it could be error-prone data hand-typed by people, or even an image from a scanned document with a mix of typewriter and handwriting.

Data governance is the set of practices and principles that professional data teams use to ensure the integrity of the information at each step.

It includes practices like data lineage, so we maximize the traceability to connect a bit of information all the way back to the raw document and the chain of custody. We describe the steps that we took to make it usable, and we archive the original raw data so it's never lost.

And data quality is another. We continuously monitor our incoming and outgoing feeds to make sure that they meet our standards. So, if a new data shipment comes in through one of our connections, and there's a material or even a subtle change in the character of that data, we want to be the first to know, so we can figure out what happened and work with the source, understand what happened, and hopefully correct that situation.

So these are just two examples, but data governance is actually a whole system of patterns and practices, and suffice to say that our data teams work to a high standard to create the most useful and trustworthy product we can make. This is why we're often the choice for journalists and law enforcement, but also for business users who just want to know they're making decisions based on the best available information.

So, two of the fun things that we get to do in our data practices are cleansing and imputation, depending on what you like to do for fun.

The raw data that we get from customs agencies often has incorrect information. Like, a major example is the weights in the U.S. maritime data that we get from CBP, the American Customs and Border Protection Agency.

The folks who complete these customs declarations at the time of loading, they have to type into fields on forms. Everybody's doing their best, but we all make typos sometimes, and there aren't extensive controls in place to make sure that declarants provide something that makes sense.

This isn't a big problem for users who are looking for particular shipments, or single pieces of information. If the number looks wrong on one record, you can work with that. But when we get into analytics, this is where we get into a real problem.

When we're aggregating records together to calculate trends, minor errors can accumulate, and they become substantial quantities that can throw off our numbers.

And then there are major errors. So, for example, if the person filling the form forgets to enter the period before the two decimals, then their weight is going to go up by a hundredfold.

So here's an example we ran into around the time of the Port of Baltimore crash, last year when the Francis Scott Key Bridge collapsed. We had customers asking to

They were asking us for help on how to see who was shipping into and from Port of Baltimore. Many saw this as an opportunity to step in and provide alternative services to the affected companies, a chance to help out and build new relationships.

And so here we've gone into our analytics section and created a search for all shipments for the year leading up to the disaster.

Now, you can see the results rolled up by the number of shipments broken down by country of origin. But if you switch from shipment count to total weight

And this is an old screenshot, which I'll explain

Then now we've got a case where, there's a big problem. Turkey is now responsible for 91% of trade through the Port of Baltimore by weight.

So this is a case where we are faithfully reporting the data as provided verbatim from U.S. Customs and Border Protection, but that data has an error. One declaration among thousands for goods from Turkey has a number that's ridiculously large. Just one person accidentally typing in 3 trillion kilograms for their weight.

So as data stewards, this is a dilemma for us. Do we report the data as provided, true to the source, or do we fix it? And if we do fix it, what do we replace it with?

So our solution is to do both. We've developed a model for identifying whether a weight is incorrect. Not just if it's 3 trillion kilograms, but actually with a fair bit of nuance, if it's more or less than it should be, using some sophisticated techniques. So we'll replace it with an informed estimate that will work in aggregate.

And for transparency, we let our users know what they're looking at. So, in the grid view, we have a note to say this data's been modified for error correction, and then in the detail view, you can always look up the original weight as reported from the source, as well as seeing the corrected value.

So with the weight corrected, we're able to see an accurate breakdown of the trade into the Port of Baltimore and the year leading up to the accident.

So here's one way that we're improving the experience and the value that our users get from the data while preserving the integrity and transparency of information.

So, another great project that we've just done is we've created a model for enhancing the imputation of TEUs. So, for those of you who are newer to the subject, TEU stands for 20-foot Equivalent Unit.

Where we're just talking about weight, TEUs are a measurement of volume. They're an industry standard, and actually the most common measurement we talk about when we're talking about how much of something.

And TEUs are not included in our most commonly used dataset of the U.S. maritime imports and exports.

So, our Lighthouse analytics application has a TEU column on the data, and this is an example of data imputation. We're not just enriching or cleansing existing data, we're actually adding new information, pulling together available signals, and using a model to generate this using our industry expertise.

So, we've had this feature for a few years now.

But we receive reports from sales and from our users that every once in a while, one shipment or another has a TEU number that doesn't look right.

So this is not a matter of going in and fixing the numbers that look wrong. We have 170 million maritime import records from the last 10 years, more than any human solution could reasonably handle. So we're going to improve things, we have to do it at the system level, at the model level.

So, there are a lot of fussy bits to worry about. You know, there are non-containerized shipments, like grain and oil, there's break bolt cargo, like machinery and steel beams, or LCL, less than container load shipments, which are only part of a container. We use the reported weight in these calculations, so you can see how our weight correction work actually pays dividends here.

And we have similar problems with bad data, like container sizes that could never exist because somebody put in an extra zero.

Some of these clues have to be pulled out of raw, typed-in text fields, not structured data, so our model does its best to parse out and interpret what was typed in there and use it when it can. So from our testing, we can see that the new model has greatly improved the accuracy of our TEU estimation.

And, in fact, our data is also comprehensive enough, that we can validate these numbers against macroeconomic or census-level data from government agencies like the Bureau of Transportation Statistics. So, when we can show that our numbers, we just add up everything in the database, and we see that it lines up to the total economic activity of the country, we can feel good that we're achieving a high level of fidelity, painting a true picture of trade across the country.

Alright, so this brings us to the next data imputation project, where we've made use of AI to do something a little more exotic.

So HS codes, for those of you new to this world, are a global standard for categorizing goods. They're extremely familiar to folks who already work in this area.

HS codes use 6 digits, 3 pairs of 2 to group goods by chapter, heading, and subheading. Each level is more specific as you go. There can be more digits after HS6, but those are specific to countries. The first 6 are a global standard.

So, HS codes are extremely useful. They're a necessary part of trade. For example, many tariff regimes have different rules for different HS codes.

When you want to search for types of goods, instead of trying to guess keywords from the description, the code is much more accurate. And when you're aggregating data to report on trends, being able to break down shipments by HS6 is an ideal way to partition data on types of goods.

And, notoriously, devastatingly, the U.S. maritime import and exports, which are among our most popular global datasets, do not publish HS codes. If you work in this area, you've already discovered this shortcoming, probably on day one.

Well, we have tackled this issue as an AI data imputation project. So, the data doesn't contain explicit HS codes, but it does have a lot of signals, clues to what the code for a given shipment might be.

One of our data science team's big projects this year was to build an AI model using a technology called machine learning, or ML. Now, these models are trained with data examples. When they get good enough, they can start producing reliable data. So in our case, we took hundreds of thousands of examples of product descriptions paired with a correct HS code, and by training the model on this data, we were able to get it working to a confidence level that we set it loose on the data.

So now, this is not an exact science. Similar to the weight corrections, we want to inform our users clearly that this is imputed data from a model, it's not raw from the source, and this is all part of our governance commitments. The data can be fussy.

So, here's an example of a bill of lading record where, oh, they typed the HS code right into the product description. So, in this case, the model's going to sniff out that there's an HS code right in there. This is the strongest signal that we have. It represents about 17% of all records. So, great. In this case, we pulled it into the product description.

And there are also cases where we'll have multiple HS codes. The U.S. maritime data is not itemized, so you can actually have multiple items under one record. Great. Okay, so we pull out two HS codes. This is why you'll see multiple values in this column separated by commas.

Now, in the next example, we get into where the model has to kick in, where we're going to use this AI to impute something.

So we have a pretty low-effort description, that's kind of awesome. But here's what it means—you know, we can probably reliably say that this is part of Chapter 18, cocoa, and 1806, chocolate and food preparations containing cocoa, but once you see the definitions for the HS6 level codes on the right, you can see we don't have enough information to say something specific. So the model's going to stop its confidence level there at 4, and this is the result that we get.

So I think we've managed… so here's a table of how this breaks down by confidence level.

I think we've managed to deliver an HS code and imputation solution for U.S. maritime data using AI that provides a lot of value to our customer. Based on the confidence level from the description, it falls back to HS4, and then for those where the description is ambiguous or unreadable, we take the L. We don't put an HS code on that one. So we have a certain level of coverage, but we're very comfortable with our accuracy.

Whether this type of solution is going to help you might depend on what you're doing. If you're aggregating data to look at trends, or to see a proportional breakdown of types of goods from a vendor or a country, this is extremely helpful in those types of analytics.

Now, if you're doing a needle-in-a-haystack type of search, you're going to want to use these HS codes in combination with other keywords and other fields to find exactly what you're looking for. Those specific records you're after might not have been categorized precisely enough by the model, but the code can still serve as an additional signal boost in your searching.

All right, now this feature leads pretty naturally into our next subject, which is one of our big new features this fall. So far, we've been talking about U.S. maritime data exclusively, and it's certainly a popular data set.

But our customers depend on dozens of other global data sets to get their info. So, for example, we just added Vietnam data last month, and we're launching comprehensive trade data for Indonesia in just a couple of weeks.

These global data sets have a lot to offer. So, for example, they can describe exports into the United States from a different perspective, and they often contain key info that's not in the U.S. data.

Another example is that across our datasets, we can collectively describe a major slice of Chinese trade activity, even though there's no trade data set being offered right now by the Chinese government.

Now, until now, adding extra datasets to your ImportGenius account meant adding more work. Users would have to go visit multiple pages in the app and build specific queries to get the collected data that they're after. With this new interface, we can offer searches that span the globe from one place.

And conveniently, one of the global search fields we're offering at launch is HS code. So, of course, because HS246 is a global standard, as we just learned.

Most of our global data sets do have published formal HS codes included, along with lots of other goodies that aren't available in the US data.

So the U.S. Maritime set now fits neatly into this system, and if, for example, you want to search for all the chocolate exports into the United States from South America, now a single code can filter across U.S. imports and our 12 Latin American data sets.

Oh, I want to make a quick note here.

You can search for these HS codes, and you see a fly to help you find the code that's right for you, if you're not experienced with these systems. This is a feature that we're shipping next week. Daniel wanted me to point this out, because he's going to be doing a demo from a production service that doesn't have this feature, so I wanted you to know this is one that's on the way, but you won't see it in the demo that's coming up.

All right, well, oh yeah, so here we're seeing our results, right? So we've managed to pull in all sorts of chocolate exports into the United States, from various countries.

All right, well, I hope I've been able to show you today how we're using AI technologies, not in such an obvious way as a chatbot, but integrating this new high-tech stuff into our backend systems to power up our data practices and add value to the core of our product, which is the data.

All right, I'm gonna hand it back over to Janine.

Jannine Krish:

Thank you so much, James. That was very… a very comprehensive feature update, and what really stood out is that they're not just technical improvements, but tools that help people make decisions with confidence.

There is one more poll I'd like to do. I realized I didn't actually share the results of the last poll, so before we jump in, here are the results.

So you can just take a quick look.

And let's quickly jump into our next one, because this is something… we are hearing a lot about.

So… what do you find most challenging when presenting supply chain strategy?

Across your leadership team, or across your organization?

So if you could take a moment to answer this, and then I'll share the results this time.

Okay, I still see some answers coming in.

Alright.

Pretty evenly, or quite distributed. Around all of these options. Thank you for taking the time.

I'd like to actually pass it off to Dan now. We're just gonna give you a live demo, and hopefully improve some of the confidence around reporting up the line to leadership and across your teams.

And just a friendly reminder, if you have any questions, please pop them into the Q&A, and feel free to book a demo if this interests you.

Alright, over to you, Dan.

Daniel Mikhail:

Thanks, Jannine. Yeah, hey, everyone! So, I'm gonna be showing everybody a demo of our ImportGenius platform today, and it's going to follow two different scenarios. So, the first scenario… actually, even before I get started on the first scenario, I'm sharing my screen here, the focus of these demos are gonna be coffee. If you were tuned in to our spring webinar, where James and I talked about AI and the initial piece of this, part one, I would say, of this webinar. I used olive oil. That was inspired by my dad. My dad loves olive oil and was very interested in where his olive oil was coming from.

For this webinar, I decided to use coffee because I feel like it's something everybody can relate to, and every time I talk to Janine or James, I see them drinking coffee. So, I chose coffee, hopefully you all like coffee, and keep in mind, as we go through these demos and these examples, you know, I'm using coffee as the example, but you can use that for jet engine parts, you could use it for…

Nuclear engines, you could look… use it for something as simple as, you know, valves, or brackets, or whatever you have in terms of a product.

or commodity. So in the first scenario, what we're going to look at is commodity import volume, a little bit of what James touched on with the TU corrections and volume corrections. We're going to show you a little bit about the analytics and what the challenges that that type of analytics solves

for companies. The biggest challenge we see right now is just understanding the volume of competitive product, or product in general, that's entering your market space. It helps affect decisions.

Across, you know, how much are we going to manufacture this year? Are our competitors bringing in more than they did last year? I want to understand that, and I find that leadership, and more and more on my demos, leadership is asking for visibility. They're asking supply chain folks, what is the forecast? What are the analytics? What is the data telling us

So that we can better advise our decision-making strategies. The solution to that is, for me, obvious, of course, but for people who haven't used trade data in the past, a little less obvious. What you need to understand is, what are the shipment volumes?

Where are the goods coming from?

What is the consistency of those goods? Are they one-time goods into the market, or is it coming on a consistent basis? And who's that importer who's bringing it in? And who's their supplier?

That paints a clear picture of the competitive space, and without further ado, and without reading this slide to you, I'm going to show you a demo.

So what we have here, if everybody can see my screen, is a line graph that's pretty flat.

And what I've done is I've created a search for coffee beans over the last 5 years.

And so what you'll see in this line graph is that the demand for coffee in gross weight in kilograms has remained pretty steady over the last 5 years, with a slight increase, and that's… I'm gonna blame James for that.

But you can also break this data down instead of yearly, but visualize it monthly. And as you go deeper into weekly and daily, it might take a second to load here. I'm gonna do the weekly, I'm not gonna do the daily. Here's the weekly.

And so this is the import volumes by weekly, and if we had daily, it would be a lot of data points over the last 5 years, but certainly it's an option to you in a shorter period of… well, you could do it in any period of time, but it'll take a second to load.

The next thing I'm gonna do, I'm gonna pop back into yearly here, and I'm gonna flip into a pie chart. So right now, we're looking at the volume gross weight coffee beans coming into the U.S.

Now, where is that coffee coming from?

And when we flip to a pie chart that analyzes country of origin by gross weight in kilograms, we can see a majority of the U.S.'s coffee is coming from Brazil, then Colombia and Vietnam and Honduras and Guatemala, and I'm not going to read all the countries, but all of the data is there available, and so you can see the top 50 countries—Djibouti, Tanzania—where we're buying coffee beans consumed in the U.S. from.

You can download these results pretty easily via PDF or Excel with a single click.

And you can either take those offline and present them to leadership as they exist in a PDF, or you can play with those Excel spreadsheets in your own CRMs or Power BI tools to tell different stories and give different visualizations, whether that's for supply chain planning or for your leadership teams.

Another cool thing that you can do with our data is you can view by importer and exporter.

So, if I just click here, and then change the field Analyze to the Importer, we want to see, okay, well, who are the largest U.S. companies who are importing the most coffee beans by volume?

So, in this case, you can see that Rothfoss Corporation is bringing the most coffee beans globally into the U.S. over the last 5 years. Here's the number of shipments, the containers, the TEUs, and the gross weight.

And then, similarly, you can do this for the suppliers. So, who are the largest foreign suppliers? Oops, I clicked into shippers. Same thing, but visually. That one was the actual data.

But visually, here are the largest suppliers of coffee into the U.S.

So regardless of what part or product you're looking for, you're gonna see that by volumes, and you can organize it by largest supplier to smallest, etc.

If you wanted to look at the particular shipments themselves, you can also click into shipments here.

There we go.

And you can see all of the individual shipments, you're gonna see who the carrier is, the foreign port, and a lot of other data that's available.

We were looking at the analytics, and we were seeing, in terms of importers, the largest importer of coffee beans into the U.S. was this Rothfoss Corporation. And you can see here that all of these companies are highlighted in blue.

So what you can actually go ahead and do is you can click on that highlighted blue, and it's actually going to pull up an overview for that company. It's going to show you their shipment history over time on the overview page. It's going to show you their U.S. address, it's going to show you their core, main industry sectors, their primary business model, their key value props, and it's gonna show you all of the sources that we pulled that from.

From there, you can view the individual shipments for just that company as a standalone, and you can click into all of their suppliers and figure out who those suppliers are, etc.

You can see any news on that company, or potential synergies between your company and theirs.

You can view all of their top suppliers by volume, which is a super powerful tool, and then get a better understanding of the countries of origin they're sourcing from. This is very important in regards to tariffs, understanding your competitor supply chains, and your own supply chains, and potentially where their risk lives.

And then if you click into a bell company, James told me this would be the most important part to show you guys, is you can see their competitors, so if you're vetting suppliers, and you can get their contact info directly using machine learning and AI and large language models. So this is all publicly available information. You can see LinkedIn profiles, you can see names, you can see titles, and for companies as well, private or public, as long as data exists, you can also see financial performance and revenue trends.

Which is fantastic.

Awesome, guys, so I'm gonna jump into case study, or I call it a case study. It's scenario number 2.

Which is what James touched on a little bit earlier, and I call it trade data, but it's global trade data, and giving our customers the ability to search all different countries and global supply chains in a single query.

Why is that important? I feel we all know that supply chains are more global now than they've ever been, and being able to visualize tariff risks, hidden relationships, second, third, fourth tiers of the supply chain are critical to planning.

You know, the solution is similar, is you're looking at verified customs trade data, you're understanding who people are selling to, you're understanding who they're buying from, the shipment volumes and timing.

And the outcome is just complete uncovering of competitive and adjacent supply chains, and understanding the full, competitive landscape, exposures, and implications for your business, which is so powerful.

I'm gonna jump right into the demo and show you a practical example. Keep in mind, again, that I am using coffee as the example, I've said this several times, but whether you call it gahua in the Middle East, or cape, or coffee, or cafe, it doesn't really matter. Our global search is going to allow you to find all of the products you're looking for in a single search.

And so, in particular, what I've done here is I've gone to datasets, and I've added datasets, and I've seen a lot of questions already about what datasets we have available and what countries. Through the lens of our datasets, you can see all 182 or 192 countries in the world.

And all of their trade relationships, but the data you're seeing are from these 55 data sets that we own.

And so you can see those countries here, I've selected them all. I'm going to submit my selection. I've already created the search, so you can see the product is coffee beans.

And I've used HS code 0901. 0901 is the coffee HS code, the 4-digit. You could go 090111, which is unroasted coffee beans, or you could go 2-1, which is the roasted coffee beans.

Or you could go 1, 2, which is decaffeinated. It doesn't really matter, you're eliminating your search down, but I've, for the purpose of this, I've just chosen just coffee beans.

And you can see that the U.S. is the largest importer of coffee, I think we all knew that. And then all of the global results will be displayed by volume. So, as there's more results.

In number of shipments, doesn't necessarily mean gross weight, but in number of shipments, you'll see all of the countries who are importing and exporting coffee beans.

And you can click into those individually and get a really good understanding of their data. You can also download all of this data, you can export your report again in Excel, and it's a single click. It's very easy to do, and you will get all of the data in these lines.

The next thing I wanted to do, I've shown you how to search our global search by a product or HS code. The next thing I'm going to show you how to do is how to search our global data by company.

And what better to show you than a global company that serves up your favorite cup of coffee? Starbucks is not a plug for Starbucks, not affiliated, but certainly might be interesting to see all of the different areas where Starbucks is importing and exporting goods from for their global coffee conglomerate.

And so I put in Starbucks as a company.

You can see in the U.S. data, they have the most imports, and you can see the Starbucks Manufacturing Corporation, or the U.S. entity of Starbucks, and exactly what they're bringing in. This isn't limited to just coffee beans. This is going to show you where they're getting their coffee cups from, their coffee machines, their makers, their equipment, etc, etc, etc. Now, we could limit this search, or add an additional filter to show you just coffee beans from Starbucks, but for the purpose of this, I'm going to show you all Starbucks products.

From there, as we click into the Peru dataset, you'll see Starbucks Corporation, Argentina, you'll see Starbucks Coffee, Argentina SRL. So our data is smart enough to understand that it's not just Starbucks, but all of the entities of Starbucks within that. So you can see Starbucks Coffee of Chile, etc, etc, etc.

And the last piece I'm going to jump into as I conclude my demo here, hopefully this has been very, visual for you guys, so that you can understand the power of global data and trade analytics that we're providing, is a visualization tool for Starbucks supply chain.

And so here, what I've done is I've visualized Starbucks' supply chain, and all of their Tier 1 suppliers.

And what I'm going to do in this particular example here is I'm gonna click into this particular Tier 1. It's a Costa Rican company that they're buying goods from.

And as I click into this Costa Rican company, I can see that they're purchasing… so this is their Tier 1, their Tier 2, they're purchasing from another company.

And then this company is ultimately buying their goods from Guatemala. So, regardless of what this company was buying or providing to Costa Rica, those goods were actually coming from

Guatemala. So, very powerful. I'd be really happy to show you your own supply chains. The question I get whenever I demo this is, can we see our own? Can we see our competitors? And the answer is absolutely yes, and it requires almost no effort, is you're putting in a company name and visualizing the supply chain within seconds. Whether that's Tier 1, Tier 2, Tier 3,

So what I'm gonna do is, Janine, if you don't mind, just drop into the chat function in this webinar right now down below, a link to book a demo. You can book a 15, 20-minute, 30-minute demo with me, and my other colleagues who are very capable here at ImportGenius, and we'll show you practical examples of your product, of your commodity, of your own supply chain and your competitors' supply chains.

where you can visually map and start to use it to make better decisions using machine learning and AI. With that, I'll turn it over to James.

For one other really super cool feature that I've been waiting for for a while, James, to you.

James Orr:

Sure, thanks, Daniel.

Right, so we like to keep folks coming back to our webinars, so I try to close with a quick note of something exciting that's coming up. This is an idea that I invented by myself that I call One Additional Neat Item.

I asked the new Seed Dream image model from Seed Labs to make an off-brand Steve Jobs that kind of looks like me.

I think it did pretty well.

So, Daniel showed a global search across 55 data sets, and that represents a real mix of languages. Some of our most valuable global data sets, like India and Vietnam, are actually in English, but many more are in their country's local languages, and I think this is an issue that actually prevents some of our users from expanding into more valuable global data.

So to help with this, we're starting to hook up large language models to assist with translation.

And I'm gonna show you what that looks like.

Daniel, Janine, can you see, the prototype right now?

Great, thank you.

All right, so here is an interactive prototype, not yet in the application, where we're simulating what it would look like to interact with a multilingual data set. So you'll notice in the top right, we're searching our Mexico import data, so our Mexico import dataset

import and export that we just launched this year. It's our largest data set ever, and it's extremely detailed, and it's all in Spanish.

So, if I picture myself as a user who doesn't speak Spanish, and I don't know a single word of it, I'm in a tough spot. I need to figure out some translations. So, if I type corn into the product field, we see a feature pops up and says, hey, this is in Spanish, do you want a little help?

And so it's going to reach out

And we're going to have a large language model interpret this and come up with some suggestions for you. Now, what's interesting here is that

you know, as a user who doesn't speak Spanish, I actually… I don't know which of these use… these words I should be using. So the system's gonna come up with definitions for me to help me understand, oh, well, you know, I'm actually searching for kernels of corn, specifically, so, maize, most widely used, okay, that looks like a good choice. And now, elote.

Oh, it looks like it refers to corn on the cob. Alright, so I might choose to leave that one out.

So, and then similarly, we can play around with things like homonyms, right? Words that are spelled the same way but have multiple meanings. So in this case, I typed in lead. The system's gonna come up and show me metal plumo, which is a Spanish word for lead, the metal.

But it's going to show me some more definitions, including debrisure, guillar, primero, which have more to do with lead, which is another interpretation of this word. So you can see, we can't just, like, blindly start translating these words. We need to interact with the user, show them the meanings, and this is where it's great to have a language model that can pop back and forth and help you build these searches.

So in this way, a user's going to be able to interact with a foreign language dataset with a lot of help, and, you know, not just Spanish, which might be fairly familiar for our American users who went and studied Spanish in high school, but,

you know, Hindi in its original lettering, Russian data that's in the Cyrillic alphabet. This is stuff that is very difficult for a lot of folks to parse if you don't know those languages, so this can help you to navigate that information.

Jannine Krish:

Thank you very much, James. That was indeed a very neat item.

I'd like to actually turn it over to the audience. We've received a lot of questions around specific country datasets.

I'm seeing, questions about, you know, do we have South African data sets, Turkey datasets, Brazilian, Latin America, just to name a few. So maybe, Dan, you can cover the regions and let people know, what kind of data sets we offer.

Daniel Makhail:

Certainly, yes. And if you go to importgenius.com as a starting point, all of our available data sets will be available to look at on the website, so if you're… if I don't cover a particular country, please feel free to go to our website. We're very transparent with what we have access to.

Now, in terms of data that we have, we have global data from every country in the world, but through the lens of the 55 data sets that we do have.

So a majority of our coverage is gonna come from, North America and LATAM, but we carry all data that is publicly available. So we… we have very diligent teams that go out, and if the data is available, we will…

capture it and display it to you. We are the market leader in that space. So, certainly, if the data is available… now, if the data doesn't exist, for an example, Chinese export data.

it doesn't exist, but you can view it through the lens of trade with the U.S. Same thing with the UAE, as a question I get asked in the Middle East. There's not a lot of data transparency, and then the smaller the country is, typically financially, the less infrastructure their government has to collect and report on this type of data. So the larger the country, the more reputable the country.

In most cases, the more data that we'll have available, but all of our data sets are available online, and you can go, and then if you still have more questions.

Feel free to connect with anybody on the sales team, and we'll answer all of your questions and show you all the data that we have available.

Jannine Krish:

Thank you. The next question is for James. This is one that we get all the time, and I think everyone would benefit from hearing the answer, and that is, where does your data come from, and is there a chance of also expanding beyond maritime shipments to air?

So if you could maybe get both those.

James Orr:

It's true, even when I tell people what my job is, usually the first thing they start asking once they

understand what we do, is where do you get all that data? And it's true, it's one of the unique values that we're able to offer as a company, is that we have over a decade of building relationships with different sources, customs agencies, and contacts

to try to get as close to the source as possible. But honestly, the answer is really, it depends, you know. In the case of U.S. maritime imports and exports, those are delivered to us digitally, daily, from U.S. Customs and Border Protection. We have a close relationship with that agency, and we receive that data with a high frequency, direct from the source.

Some come from vendors who serve as middlemen in between us and other countries, and some come from interesting connections that bring us into countries that are a bit farther away and separated from us, including Eastern European nations that can shine a light into Russian trade.

So yeah, I suppose the big answer is it really depends.

Now, as far as different modes of transport go, it's true that our U.S. imports and exports provided by CBP are maritime only.

Although we're working on that.

But a lot of our other data sets contain all modes of transit. So, for example, our Mexico dataset that we just launched this year, which includes all Mexican imports and exports with the United States, it includes all modes. So, rail, truck, air, sea, all the good stuff.

So, yeah, I think if you expand into some of our global data sets, you can get a much bigger picture of different modes of transport.

Jannine Krish:

Thank you very much. Next question, this one is a quick one. The visual map that you showed, Dan.

One audience member mentioned he could see thousands of shipments, but what is the date range? Last, you know, is it a day? Is it last 12 months? How can you filter that around dates?

Daniel Mikhail:

So the visualization of the visual map that I showed is from 2006 to present.

is the data that is captured there. However, it is filtered by, volume first. So, on the first page, the relationships that you're gonna see is the relationships by volume, and as you click through the different pages.

you'll see the relationships size decrease… decreases. Now, if you wanted to view relationships via a date range, you can do that very easily just by setting your date range, and then viewing all of the shippers or suppliers for a particular company.

So the answer is, you know, in my visual map, it's data from 2006 by volume, but you can certainly see that in our iScan platform, just by searching the company name, and then aggregating and viewing all of their suppliers by date range.

Jannine Krish:

Thank you!

Daniel Mikhail:

My pleasure.

Jannine Krish:

HS code question for James.

When the HS Code Search feature launches, is this going to be available to view on all shipment data?

James Orr:

Sure, so the HS Code feature that I spoke about is really our solution for the U.S. maritime import and export data sets, specifically. That is a dataset that it… where the HS Code information is painfully absent, and because it's such a popular dataset with our customers, that was a big target for this project.

Most of our international data sets do have HS codes published by the source agencies, so we don't have to get too fancy with those. The information's already available. I hope that helps.

Jannine Krish:

Thank you. Another customer said, maybe Dan, you can answer this one. Sorry, not a customer, an audience member.

says, I use Pangea, can you please explain the difference between ImportGenius and Panjiva? I know that's a question you get all the time.

Daniel Mikhail:

Yeah, absolutely, yes, I do. And so the core difference between ImportGenius and Panjiva is how we break up and sell the data.

Panjiva, I believe, sells the data globally, so it's all global data, regardless of what your data need is. Whereas if, for us, you're only interested in imports to the USA, we can sell you just the imports to the USA, and, you know, you're having a significant cost savings.

I also think our platform and infrastructure is significantly different.

And, you know, in my opinion, obviously biased better. And then the last but not least will be the support. The support that you're getting, and the team is based here in the U.S,

Whereas, you know, for some other companies, their support teams are based globally, or trade data is a small portion of what they do. So S&P Panjiva, for example, owned by S&P, you know, trade data is, you know, less than 5% of their overall business, whereas

trade data is our business, and we take it very seriously. Our CEO… there was a question asked about, you know, can we get air data? Our CEO is sitting helping passing a bipartisan bill in Congress right now called the Supply Chain Modernization Act that will help businesses get access to U.S. air freight data.

Every person who works here has a supply chain background and supply chain expertise, and it's a big, big part of, you know, who… our identity as a… as a… as a data… trade data provider.

Oh, I'd like to bump in for one sec. I think Joshua's question about HS Code Search.

James Orr:

Might have been…

having to do with that… that feature that I showed, Daniel, that isn't quite released yet, where you type into the HS code, it actually lets you search through them by keyword or by code. Joshua, when that launches, it will be on all shipment data, so not just U.S. Maritime, but any data set that has HS codes.

When you start typing one in, you'll be able to interact with that search to pick codes.

Daniel Mikhail:

And I saw another question here. Sorry, Jannine, I know you're moderating, and James and I are jumping in, but I saw a question here about 24-hour support hotline.

Absolutely. You can reach out to us at any time via the chat function on the ImportGenius website, and I think you will all be very pleasantly surprised with the response time. If everybody does it right now, it might be a little bit slower, but certainly come back whenever you want, and randomly use our chat feature, and see how quickly you get a response.

Of a real person, not a chatbot.

But a real supply chain expert who will help you solve your challenges, which is a big, big piece to this.

Jannine Krish:

Thank you. I do have a question for James.

What… Is AI's role in all of this? Can you be more specific?

James Orr:

Oh, well, I mean, if you, if you've been tracking this technology since the ChatGPT revolution that started in 2022,

The product name is actually very prescient. GPT stands for General Purpose Technology. Part of the fun of this technology revolution is that large language models and transformer models that have

Created all these new tools and technologies.

they can be used in an endless variety of ways. Some are ill-advised, some are very creative, but the number of ways that this new tech gets used as a general purpose tool only keeps growing. So, for example, a lot of

AI features that you see popping up for companies these days take the form of a chatbot, and that's becoming a little bit boring. Whereas today, we were able to show that we were using some AI technologies in our backend data infrastructure in extremely technical contexts in ways that could increase our data quality and trust.

So…

really, as a company, we're trying to encourage bottom-up AI adoption by encouraging all of our employees to make use of new tools and try new things as a way to grow their own skills, as a way to improve the productivity of our company. And from the top down.

We're just immersed in all the new stuff that keeps coming out, running research projects, trying proofs of concept, and seeing what's really going to help our customers.

Jannine Krish:

Well, thank you all so much for popping your questions in the chat. We are nearing the top of the hour, so unfortunately, we're actually in overtime, because I did say this would be closer to 45-15 minutes, but you're asking such great questions, it was hard to stop.

So, just want to thank you again. If you didn't get to ask a burning question.

Feel free to book a time slot with us. We popped the demo link in the chat a few times.

You're also welcome to connect with us on LinkedIn or ImportGenius on our corporate account. You can ask questions there as well, and we will do our best to get back to you in a very timely manner. With that, I'd also like to thank James and Dan.

And I wish you all a wonderful rest of the day.

Bye.

James Orr:

Thanks, everyone.

Daniel Mikhail:

Thanks, everyone!

Jannine Krish:

Thank you.