Okay, good morning, good afternoon, good evening everyone. Um, thanks for joining our interactive webinar. I really have the pleasure today to welcome Mark Devil from Spire Maritime. Welcome, Mark. Hi, Bart, thank you for having me. Good morning, everybody. Okay, so before we start, it will be an interactive interview-like webinar. I want to tell you that the webinar is recorded, so if you want, you can rewatch it later. We will be taking questions, and I hope we have plenty of time for questions, but please put them in the

Q&A. Um, so we will take them at the end of the session. To keep it interactive, we will also have a number of polls. It would be really nice if you could answer the poll, as it can help us in guiding the remainder of the webinar. Um, so thank you all for joining again. My name is Bart, I'm the CTO at XYZT, and Mark Devil is here from Spire Maritime. Before introducing Mark, let me explain why we are here. Well, this is basically why we are here, and some of you might be very surprised to see a black screen, but

that's actually what is still happening a lot in many organizations. Many organizations are still sailing blind, and what Mark and I will try to convince you of is that you don't have to sail blind these days anymore. Um, indeed, what if we could really see not only what is happening now but also what happened in the past, and maybe even predict what will happen in the future? Wouldn't this be nice? And that's really what we will be talking about today in this interview. Um, so we are XYZT; we provide a big data analytics

platform specifically focused on spatial-temporal data. As you know, in the maritime domain, there's a lot of spatial-temporal data, and one of the very valuable data sources is the data from Spire Maritime. That's what you see here in the picture: you see a density map of SP AIS data in one of our sample projects that you can explore in our platform. Um, so what is XYZT? It's a no-code interactive visual analytics platform for any analyst to extract real-time but also historical insights, and today we will also show forward-looking

insights for maritime and any other data. We really try to keep it as easy as possible and as efficient as possible for you to do so. Uh, with that being said, I would like to introduce Mark. Um, I know Mark for a long time, and I also know Mark you've been in the maritime ecosystem for quite a long time already, so maybe you can introduce yourself and explain a little bit about what you do at Spire Maritime, what your team does, and basically what you will be talking about today. Sure, thank you, Bart. Hi, everybody.

Uh, yes, just to get the awkward fact out of the way first: I have been in maritime data for a long time—it's 30 years this year, so clearly I started very, very young. But it's been a very interesting journey, and as Bart said, you know, at the beginning of the ship tracking services story, people were in the dark; information was very slow to come in. So, I am the Data Operations Director of Spire Maritime. Spire Maritime is the leading provider of vessel tracking data and information services around vessel tracking and maritime monitoring. I've been with

Spire for six years and we believe we provide the unrivaled coverage of vessel positions and movement tracking, and the information that you can build on top of that. So we deliver the most accurate and comprehensive data, as you see on the slide here, and I can illustrate that through one of those slides. So I would have lost the share screen. Yeah, while you are switching the presentation, Mark, you have been using XYZT for quite a long time already. In fact, I think you were one of our first customers, if not the first, for which we're very grateful. So maybe while showing your data, you can explain why you use XYZ. Um, what's my screen displaying? Ah, so I didn't hit the confirmation, apologies. Yes, so we started using XYZT in 2020. I think we realized in recent discussions, and I can't remember how the conversation started, but we realized it was a tool for which we had a need. We had a very large data set on

which we had to do on demand. Here, because I don't see the screen, I see the Zoom app at this point. So here we go. Is that better? There you go, that's better. Yes, right. So we started using XYZT for exactly the reasons of its basic capabilities. We had a very large data set and we often needed to do our own analysis and investigations on our data set, including visual analysis for particular coverage or where a vessel's journey would have had good coverage or bad coverage. It was very geospatially oriented; we could analyze

the data numerically, but we couldn't analyze it visually. The other tools we were using were very restricted in the volumes of data they could display, and what we discovered was that XYZT had a very easy-to-use no-code tool, as Bart said, where we could load in a complete day or week's worth of AIS data, which is hundreds of billions of records, and then we could analyze it, display it, and zoom into it, and investigate it and play around with it in the XY platform. So what you see here is an example of how we are now using XY

and have been for the last four years. This is one day's worth of AIS data globally displayed on our map. There's probably about 500 million records from this one day here, because this is all of our partner sources and all of our satellite data, and we can very easily investigate this in the XY platform. So that's why we started using the tool; because it solved a real problem for us and it enabled us to perform much more detailed analytics, which we use as both a pre-sales tool and a data quality tool, and as an analytical tool. So

for us, it's been a very, very useful platform, and we use it in a small team on a daily basis. Okay, and so what we're looking at basically, I can imagine, is the historical data. Is it something that, yes, people can buy from you, this type of data? Absolutely, yes. So we sell our AIS tracking data through APIs or as historical dumps of data. This is one day's data from yesterday, so we don't have a live feed into XYZT; we have a daily update to allow us to do analysis on the historical data as it builds up. Okay, but this is data that you

can access by purchasing the Spire Maritime APIs. Our Maritime 2 GRQ API is the best example where you can request on demand the latest information for either the global fleet, a specific fleet, a specific list of vessels, or a specific area. You could get the most recent reports from all the vessels that meet your criteria at the frequency with which you pull data. So you could say, 'I want global updates every 5 minutes,' and you could pull the information, which would then display on a map like this, or you could do it on demand. Say, 'Okay, where is this fleet now?' or 'Where is this fleet?' or 'Which ships are heading to this port?' You can call our data through the GRQ API in very imaginative ways to answer the questions that you have. What you get is you start to build up a data set that looks like this: the ships' positions kind of globally or within the area that you're interested in. And if I zoom in, you can see there's quite a lot of detail. What you can do in XIST is scroll through this over time as well, so maybe it's worth us quickly showing that.

I can imagine many people, or at least people in the maritime domain, are starting to use historical data for machine learning purposes. How far back can you go with the data? Is it more than a year that you keep the data? Absolutely! We don't throw data away; we realize the value of data, especially when you're driving machine learning and AI engines. The larger your historical source set, the more accurate your predictions are. Our archive goes back 14 years. Our satellite archive goes back 12 years, and all of that data is available to our customers if you purchase the historical access products. If you purchase the live product, then you see the most recent positions, or if you purchase the raw product, you see the last 30 days as a continuing change in window. But yes, we have 14 years of archive data for global AIS. Very cool! And for the real-time data, how real-time is real-time? I can imagine many people ask you this question.

Absolutely! Real-time is sub one minute for both satellite and terrestrial-based AIS. For our dynamic vessel-based AIS, the latency is a little bit higher; it's kind of between 7 and 15 minutes, so it's near real-time for almost everything. Okay, that's really cool. But today, Mark, I think the key topic is to talk about some new data products that you have introduced into the market. They're not looking backwards in time, not looking at the current state, but actually looking forward in time. Can you tell us a little bit about what those two data products are? Absolutely! So we have two. Our first one is what we call our vessel support predicted ETA. It's a little bit of a mouthful, but it is what the name says. This is for every vessel that we're seeing reported in AIS. The major vessels with IMO numbers and the cargo-carrying vessels—we predict the destination port for those vessels and therefore the route they would take to those ports, and the ETA when they would arrive at those ports. Now that information is partly user-transmitted, as many vessels will say, 'This is my destination, this is my ETA.' What our

system does is enhance that information by predicting the routes and arrival times based on historical data and current conditions. This allows for better planning and operational efficiency for shipping companies and port authorities.

Predictions validate or improve what's being reported by the vessels themselves, and we find that the ETAs reported by the vessels are quite often not as accurate as they could be. They change as they get closer to the port, but our predictions of when the vessels will actually arrive are slightly more accurate than those declared by the vessels. Also, if the vessel is not declaring its destination, we predict where we think it's going by looking at the type of ship, its trading patterns, and history.

It's gone, where it could go, and maybe what is reported, converting a variation of the name into a very clear prediction of where the vessel is going. This gives us a forward-looking view of things. As you and I discussed a little while back, when I asked, can we plot in XYZ future positions? What I'm showing you here is data taken this morning of vessels with their predicted ETAs and predicted routes. I've just shown a small example of the routes to illustrate, and this is

But if we expand this out, you can see these are the routes that we're predicting for these vessels. Now, I filtered this down to a very specific set of data. I'm sorry, this is the wrong window. This is the on-demand routing; these are the predicted routes and predicted ETAs. So these are the vessels where we're predicting the ETAs and the routes. We're updating that every 15 minutes. If we see a variation, we will refresh the prediction for a vessel; otherwise, we will regenerate it again twice a day. So for these vessels, I think I filtered it here for container ships, so there are a lot of these vessels, and we're predicting where they're going. Now, many of them will be declaring it, while some are not declaring it, and we are predicting variations on what they declared because it hasn't been updated. But what you can see again is the future routes. They start to expand the timeline of the data that we're displaying. Oops, wrong slider. This one here opens out, and you can see the vessels moving along their predicted routes, and then they all

slowly arrive at their ports. What you can see is where they are going to be in the future. This is the current and a couple of hours ahead situation, and if I slide that along, you start to see where vessels are going to be. From an operational perspective, particularly from a port viewpoint, you can see when the vessels which are expected are going to be arriving in your port. If you're looking at it from a market situation, you can assess the cargoes on these vessels. For instance, if you're looking at crude oil, you can

analyze each port by the type and how many vessels are waiting at each time. If you combine that with the prediction of when other vessels are coming, you can start to get a real performance planning type operational view of things. You can say, okay, this is the turnaround time, this is how many vessels I had in port, and now this is how many are coming. You can start to see whether vessels are going to have to wait or whether they're going to be able to go into port. If there's a waiting situation, you can

start to plan around that. You can slow down the vessels, and as a port operator, you can start to contact vessels and rearrange the schedule for each vessel depending on when you think they're coming. If you're a port operator, you can say, okay, these are the vessels due to arrive today. Are they going to arrive today, or are they going to be late? In which case, I can then reallocate their slot to somebody else and contact the vessel to reschedule them. So there's a lot of operational planning involved.

type decisions that could be made using this information. So we've gone from the now situation to the future situation. Sorry to interrupt here, but this brings to mind a question that I, as a more tech-oriented person, have: because you are predicting the future and because time advances, the future today is not the same as the future tomorrow. You can sort of verify your prediction—how accurate it was compared to the real world. Did you do that exercise? We did. Our data analysis team used the predictions that we generated.

We captured the history of those predictions and then compared them to the actual arrivals of the vessels to see how accurate our predictions were. Generally, it was about 10% more accurate than the predictions transmitted by the vessels. So we were measuring it against the time when we recognized the vessel arriving in a port. We had a set of port polygons through which we created events and recognized arrivals and departures in port, and we measured our recognition of the vessels arriving in the port against

those which had been predicted as the ETAs in a vessel's AIS versus our own predictions, and we found our own predictions to be more accurate. They especially became more accurate in the last five days approaching the port, and it became more accurate each day as you got closer to the arrival as well. So we're very confident that it's an accurate and correct prediction that we're generating. I understood also that your algorithms, or your predictions, adapt to the changing situations. So maybe not so much

for container vessels, but maybe for tankers. I can imagine they change course quite a bit, so you take that into account and you update the prediction. Yes, so the engine that generates these predictions is running every 15 minutes and looks for a deviation from the previously predicted route. If a vessel is observed to have deviated from its route, then we regenerate the prediction again. If that deviation is something like 20 nautical miles off from the B route, then it will regenerate the prediction. So

we're looking to refresh at a fairly high frequency so that we can continually get up-to-date information. That's really cool. And this is also accessible through an API? This entire thing? Yep, so this is a standard feature of our Maritime 2 Graph Q API, where you can select the vessel's predicted ETA feature, and it will become part of your standard response that you can request through the API. So as well as getting the vessel's latest position and latest voyage information, you can get the predicted voyage

information as well as an add-on feature. Okay, super cool. Thanks, Mark. Maybe before going to the second predictive product, it's a good time to do a small poll question. So you will see the poll popping up, and there are a number of options. But the question first: what are your biggest challenges in maritime operations today? First option: lack of easy-to-use analytics tools. Second option: lack of accurate predictions of destination and vessel ETA. Third option: insufficient integration with existing workflows and

Then the final and fourth option is data gaps in high traffic zones. Um, please submit your answers; it will take a minute to give you some time. I see still answers coming in, so we'll wait a few more seconds. All right, I think we are about there, so let's have a look at the result of the poll. Okay, it's a bit uniformly spread. I think the winner is insufficient integration with existing workflows, which is quite an interesting one. I think APIs can help a lot with that integration, but indeed it's a bit.

Spread all over the different challenges. Um, thank you for providing your input. And that brings me, Mark, I think, to the second predictive product, which is called the routing. Absolutely. So our second predictive product is the routing product. Now, that sounds like it might be the same; it's not quite the same because this is an on-demand routing product. So what you can do is you can make a request to our routing API, another GraphQL service, where you can say if I want a vessel to go from point A to point B or

port A to port B, or point A to point B, what would the route be? How long would it take? And you can put in the normal kind of voyage management criteria for routes: things like I want it to go through the big canals or not, I want it to avoid the piracy areas or not, I want it to avoid the emission control areas or not, all these variables. And then you can say what would the route be? The classic example for that, in my experience coming from a ship broking background temporarily, was when you have a ship

broken situation and you want to charter a vessel to pick up a cargo. You look at the available vessels in the market and you say, okay, which ones could get to my port in the window I have to pick up the cargo? What I've done here on screen is I called our API, and this is for gas carriers. I said from the current positions of gas carriers, these are the ones where we didn't have a clearly predicted destination. Which ones could make it to Singapore? So give me the route for all the vessels in the current positions to Singapore, and then that

gives you a prediction of when those vessels could be there, and you could select which ones would make it within the time scale that you're interested in. Then also, you'd have commercial arrangements and discussions with those vessels. So it's very much a kind of THL looking planning exercise. For each of these vessels, we asked the question; we had to make an individual call saying, okay, when would this vessel be able to arrive in Singapore? What would the route be? What would the ETAB be? And you can see on the

map where the vessels are, and you can see as you scroll through the time when they would be arriving in Singapore. So again, as we did just now, you're looking forward, okay, this following the predicted routes when the vessels will arrive. You can see that most of them arrive within the next two weeks, and you can use that information to then do the next set of investigations or planning or commercial discussions that you need. So it's very much an on-demand kind of THL looking prediction utility which you can call.

through an API, I guess, because it's again, yep, it's a great API. You simply say this vessel from here to here, give me the route and the ETA. That's really cool. Um, yeah, that brings me maybe to what are like these two new products that you brought to the market. Um, you already mentioned a little bit the use cases in chartering and so on. Do you see other use cases, um, where these products can really add value? Absolutely. Anything that needs to plan and look forward in terms of where vessels could be, where vessels are, where they're going, and where they will be.

be at a particular point in time. Um, maybe an example would be if there's a storm in a certain area. Okay, let's look at the vessels on their known routes and let's just go forward through those routes and say, okay, who's going to be in this area at that time? Who's going to need to divert? Who's going to need to change their plans? What risks have I got? Um, and if there's like a major incident, a tsunami or an earthquake or something worldwide, normally you would look at vessels, where are they? Now, what is my effective fleet? You can now

look forward and say, okay, which vessels are due to come into this area? Uh, which of those operations are going to be affected? So anything where you need to plan around a vessel's location over time, this kind of really feeds into that decision-making. Okay, yeah, I was also thinking about maybe like crew allocation, like, uh, I don't know, for maintenance or for inspections or absolutely. We've had customers who have wanted to look at vessels coming into their service areas or vessels planning, um, service activity. Okay, where

will they be? So then you can plan, you can schedule to have them met in the right port. Um, I've worked with companies in the past who needed to plan to put people on the dockside to meet vessels, and they had to plan where those vessels were going to be to get the people in the right place to meet those vessels. There are lots and lots of different situations. Okay, I see. Mark, we also have a lot of questions coming in in the chat, so maybe, um, what we can do is I can move on, um, maybe to the next poll and then, um, I have a little fa

vor to ask from the audience, um, and then maybe we can go into those questions because I can imagine the questions from the audience are very, very relevant, of course. Okay, so I think, um, let us get up the second poll. So what is your primary interest in maritime data analytics? It's A, enhancing real-time decision-making with live data, so this is talking about the real-time data; B, conducting deep historical trend analysis, so this is talking again about historical data and the analysis; C, including navigational and vessel

factors to generate shipping routes; and D, the fourth and final option is evaluating analytics platforms for better data integration. Um, so please highlight the one that is most appropriate to your case. Um, we'll wait a few seconds again for your [Music] input. Okay, here we are, so a lot of real-time decision-making with live data, and I think predictive data is enhancing the real-time decision-making, of course, and influencing. So that's a really cool, uh, cool statistic to see. Um, so that being said, let me take over the screen for a minute, Mark, but I

I might expect you to take it over later again if you get many questions, um, so I'm going to share my desktop. I have to ask one more favor from the audience. A couple of weeks ago, we introduced a new feature in our platform called 'Just Ask XYZT.' Just to illustrate what it is before I call on the audience, I'm going to demonstrate it with a little demo using Spire Maritime data. Let me see if I can get my demo ready; my screen was not so responsive, so I'll just load it up.

Again, due to the screen sharing here, I have a number of projects. Let me open one that I prepared for the webinar, and actually, I had prepared it for the launch of that new feature. This is historical Spire Maritime data in the platform. It's quite old; it's from four years back, but it's just an illustrative sample in the platform. You see the traffic here in the North Sea, which is quite busy. What a lot of our users do is filter the data, for example,

only certain container vessels like you did, Mark, or tankers or whatnot. Instead of having to set filters manually by clicking around or by coding a little filter, we started using Gen like many do. With Gen, you can do a lot, but we also know Gen has its quirks sometimes, and that's why we are still in beta. Today, I want to ask the audience who might be interested in helping us test the beta and achieve better results in the end. Already, it's extremely powerful and helpful. For example, I can now ask questions

like, um, 'Show me all traffic.' I'm doing it here in this little filter text box, from the Port of Antwerp to the Port of Rotterdam. Then we will use Gen to generate a filter. As you can see here, it generated a filter, which is like a sequential filter of conditions. You can do it in the spatial and temporal domain or in the attribute domain. I'll do it in the spatial and temporal here, for example, by adding a condition of less than 5 days. This way, you can start drilling down into the data in a very intuitive way. What you see here

now in the end, let me choose a slightly different visualization. You can start drilling down into the data just by asking, so it's super intuitive and super efficient to do. It uses a large language model in the end. Just like Mark, you are using AI or machine learning to predict, we are also using machine learning and AI to improve user interaction and make it even better. My call to the audience is that we are still looking for beta customers. Do you want to join the

beta program? If so, please let us know. I think we have a small poll coming up that asks you the question if you want to join, or you can scan the QR code on the screen. It would be really great to have some users in different use cases. We're all in maritime here, but it might be interesting to have some people from, I don't know, the Port Authorities, or maybe some vessel operators, or perhaps some intelligence analysts, because it can really give us insights on what to improve and what

to add, so I'll wait a few seconds until the answers come in. Okay, thank you all for providing that. I think before going into the Q&A, um, the message today was not only to talk about data and analytics but really to try to make steps towards organizational operational cases. We really believe maritime organizations should not sell blind. We all know there's real-time and historic data, and now there's also great quality predictive data. We, ourselves at XYZT, are in the business to make sure that

people can use the data. The challenges are that the data is big, and it also has a lot of dimensions, not only spatial and temporal but typically also all the attributes and so on. Many platforms have challenges handling that. What we see is that AI is not just a buzzword; it's really transforming many verticals, many domains, and the maritime domain is being transformed with AI. You can see it happen in many different areas, such as predictive data, user interaction, anomaly detection, and so

on and so forth. Um, with that, I would like to thank Mark for your time and preparation, and for being a longtime customer as well. I would also like to thank the audience, of course, for attending in such large numbers. With that, maybe I can have a look at the Q&A and the chat box as well because I see questions coming in on both. So, yeah, the question about how far back historical AIS data goes—I think Mark you answered that: 14 years, yes, 14 years. We have an archive of 14 years of data. Okay, um, next

question about the update rate of real-time AIS data: how much does the distance from terrestrial stations and the position of satellite constellations impact it? Uh, it does impact it. The satellites are much better at receiving ships' positions once they are outside of port areas. So the two work together: the terrestrial-based port land-based receivers receive vessels that are within their area of reception. Once the vessels move outside of those areas into less congested shipping zones, then the satellites pick them up much more

reliably. But we also have vessel-based AIS, which also picks up vessels and transmits what they receive, so the three together give us complete coverage. Yeah, and really one of the key USPs, I would say, of the SP data is that it's just global; it's everywhere. Yeah, okay, thank you Mark. Um, another question I see is that historical AIS is great. So many startups are implementing data analytics and raising money. Are we at the peak of AIS popularity right now, or are we at the start? What are your forecasts? We're clearly not at the start.

We've been selling AIS data for 15 years as an industry, and AIS has a story of continual growth. So I think we're not at the peak yet. More and more companies are learning how to use business intelligence to base operations around data-driven decision-making, and I think particularly within maritime, there's still a long way to go before we get anywhere near the peak. If you then add in the AI aspect of this as well, the machine learning and AI-driven decision-making aspect, it's an upward path. There's a

long, long way that we can take this. True, yeah, I could not agree more. Um, indeed, if you look at, if we talk a lot with organizations like shipping companies or port authorities as well, and indeed, even though the industry has been using the data for so many years, I think we're still at the start of optimizing and making everything more efficient and, uh, more intelligent using all the data sources that are available. Um, let's take a few more questions. There are many questions, so I apologize if we cannot answer all of them.

But we will follow up through email for sure. Um, next question is, does the platform support easily fitting data into some data model and is it ready for predictions? Um, for example, fitting the AIS data into a random forest and doing predictions with the UI? Um, so that might be a question for XYZT, but I'll answer that quickly. The XYZT platform by itself does not do predictions on the data. Um, so in this case, the predictions are done by Spire Maritime and Spire, you are selling those predicted data products. Yes, yes, and as I said, you can access

those predictions through an API with a set of filters, um, so you can narrow down what it is you're asking for as well. So you make your own request for what you want through the APIs. Yeah, um, and I think both of our organizations can, of course, help with any custom type of analytics if that is needed. Um, I see a question here: can XY.

assist in analyzing historic movements of vessels to identify dark trade, such as unusual STDs, manipulation, etc.? Um, yeah, the short answer is yes, of course. Um, and it's one of the key use cases why people use the XYZ platform. It's used in many cases, for instance, also, um, in Nigeria, where there's a lot of oil theft, unfortunately, for the country. Um, and there, indeed, by basically cross-checking AIS data with port reports, authorities could recover a lot of lost taxes. Um, question for Spire for you, Mark: does Spire use ocean carrier container

vessel schedules as a data source for your predictions in addition to the historical data? Not at this point in time, no, we don't. Okay, so maybe something for the future. Um, let me see, let's take a couple more questions and then we can probably round up. Um, how many satellites does Spire have? This is a question you know that, Mark, can you still follow up correctly? I do not know the exact number of our AIS-enabled satellites we have. Um, the ability to swap satellites in and out of service—we had two constellations. We had the

real-time satellite constellation, which is pretty much fixed at around, uh, just under 60 satellites collecting the real-time satellite data every day. The number in the second constellation changes, and I don't have the exact number to hand right now. Um, with, I mean, Spire as a company has launched over 150 satellites in the last 12 years, uh, and they have had a fortuitous lifetime generated, so they come in and out of service, but I don't have the exact number to hand. We focus more on the volume of data that's being

generated as part of the service. Okay, cool. And maybe an anecdotal question: is it true that satellites are named after employees at Spire? Yes, there's one up there called Lio Deil. We each get to name a satellite. So once you join the company, you get onto the list, and when you reach the point where a satellite is launched, you get to name it. That's really cool; it's a nice little feature that might attract some people to the company. That's really cool. Uh, let's do two more questions. So, uh, question maybe for both of us, but I'll

start with you, Mark. Um, how do you see the increased connectivity impact this whole Maritime awareness? Well, increased connectivity is perfect because decision-making is all about how quickly you can get information and how quickly you can use that information. So, um, particularly with our vessel-based system, where it's using satellite communications to transmit the vessel data back to us, we're getting that data there quickly. So we have real-time data within one minute from the real-time satellites. We had the

vessel-based data within kind of six to seven minutes. It really enables the quick processing of data and therefore quick decision-making. I think it's critical. In the past, prior to this, data was coming in on a daily basis and it was much slower. Now we have the ability to make real-time decisions with real-time data. Okay, yeah, I think I totally agree, Mark. For us, it's the same thing. I think the increased connectivity at sea is changing the maritime domain. Um, but we also see, aside from AIS data, we see a

lot of sensor data being transmitted and collected, and also even environmental data on the current state of the ocean becoming available. Um, and yeah, we have customers that ingest that data together with AIS data into our platform to be able to analyze all these parameters as well. Um, and then let's take a final question. Let me see here. Um, what about, yeah, this question: so what if there is no AIS data? What if the vessel turns off their AIS? I guess that's the question. That's a very good, very good question. We have a

partial solution for that. Okay, um, we are using radio frequency geolocation data to recognize transmissions of AIS messages without a GPS position. So vessels will often try to hide their location by turning off the GPS connection or by spoofing the position information within AIS, and we can detect when that's happening through the radio frequency geolocation. We can detect where the transmission came from and therefore whether that's from a vessel that's saying it's somewhere else or whether a vessel is

saying, 'I don't have a position.' And we can tell that not globally; it's in the early phases of product development. But certainly for vessels at sea, we can pick that information up, and through the radio signals received by the satellites, we can identify such things. If the vessel goes completely dark, then there are other services being worked on to try and correlate ships that could be in reported positions. Other companies are correlating recent AIS data with imagery and things like that, satellite aperture.

radar imagery and things like that. So there are services developing, but I don't think there's a complete service yet. But we're part of that solution going forward. Yeah, there's a lot of interesting activity in that domain, and in the end, it's also spatial-temporal data; it's about location and time, and some attributes maybe. We've also seen this data in our platform already, Mark, and we've also seen SAR data from satellites, right, collected to do these types of analysis. So definitely, for sure, if there's no AIS data, it

can be an interesting sign to investigate. Um, indeed, gaps can be of course seen from the AIS, and one with AIS observation is the question: okay, which vessel could it have been? You can use the routing prediction to say, okay, which vessel from their last reports could have gotten there? It could have been this vessel. So I would say, oh, I've got maybe a 250-meter-long vessel; maybe you can tell it's a tanker or a bulker. Which bulkers could have gotten there? Yeah, that's really cool. So it sort of reduces the search space, exactly.

Searching for a needle in the haystack anymore. Okay, cool. I think we took our time; we handled a lot of questions. For those that we didn't handle, we will follow up through email. So to conclude, I would like to first thank Mark. Mark has been really dedicated to preparing this interview very well, and it's a real pleasure to have this vast expertise shared with all of us, I think. I hope we all learned something today; at least I did. And with that, I would also like to thank you all for joining the webinar.

And of course, it has been recorded, so you can always rewatch it or share it with your colleagues. So with that, thank you, Mark. Thank you, everyone, for joining. Thank you.

Navigating From History to Predictability: Leveraging AIS for Maritime Awareness

정영배

스파이어 마린의 마크 데블과 함께하는 대화형 웨비나.

웨비나는 나중에 볼 수 있도록 녹화됩니다.

마지막에 Q&A 세션과 상호작용을 위한 설문조사를 진행합니다.

조직의 '맹목적인 항해'라는 과제를 극복하는 데 집중하세요.

공간-시간 데이터를 위한 빅데이터 분석 플랫폼인 XYZT를 소개합니다.

마크 데블이 자신과 스파이어 마린을 소개합니다.

선박 추적 데이터의 선도적 제공업체로서 Spire Maritime의 역할.

Spire는 2020년부터 데이터 분석을 위해 XYZT를 사용하고 있습니다.

대규모 데이터 세트 및 시각적 분석을 처리하는 XYZT의 기능.

API 및 기록 덤프를 통한 Spire의 데이터 액세스.

스파이어의 예측 데이터 제품 소개

선박 지원 예상 도착 예정 시간: 목적지 항구, 경로 및 도착 예정 시간을 예측합니다.

선박이 보고한 도착 예정 시간과 비교한 예측의 정확도.

예상 경로 및 도착 예정 시간을 XYZT로 시각화합니다.

예측 데이터를 사용한 운영 계획 및 의사 결정.

실제 도착과 비교한 예측의 정확도 평가.

선박 편차에 따른 예측 업데이트 빈도.

Maritime 2 그래프 Q API를 통해 예상 도착 시간에 액세스합니다.

다양한 기준에 따라 경로와 도착 예정 시간을 예측하는 온디맨드 라우팅 제품입니다.

사용 사례: 화물 픽업 기간에 맞춰 선박을 용선합니다.

예: 싱가포르에 도착하는 가스 운송업체의 도착 시간 예측.

예상 경로 및 도착 시간을 XYZT로 시각화합니다.

온디맨드 라우팅 예측을 위한 API 액세스.

전세를 넘어 예측 데이터 제품의 광범위한 적용.

사용 사례: 폭풍 피해 방지, 사고 대응, 승무원 배정.

예: 선박 서비스 및 승무원 회의 계획.

해양 데이터 분석에 대한 주요 관심사에 대한 최종 설문조사.

'Just Ask XYZT' 기능 소개 및 데모.

'Just Ask XYZT' 기능을 테스트할 베타 고객을 모집합니다.

주요 요점 및 마무리 발언 요약.

해양 영역에서 AI의 혁신적 역할에 대한 논의.

청중의 질문에 답하는 Q&A 세션.

Navigating From History to Predictability: Leveraging AIS for Maritime Awareness

Navigating From History to Predictability: Leveraging AIS for Maritime Awareness

스크립트