AI has brought winds of change into various sectors, and the maritime industry is no exception. Although quite traditional at its core, it is slowly embracing AI and ML tools to solve its main issues. Rory Proud, today’s guest of Jarek Jarzębowski, has been observing changes in the sector for years, witnessing the technological boost and actively contributing to its progress. In his conversation with our host, he breaks down the main challenges the maritime industry faces and their scale.
Whether you have some experience in this field or are a complete rookie, this topic is fascinating to plunge into. From how vessels are trying to combine data tracking with their own best interests to deceptive shipping practices and their prevention, we’ve got you covered. Discover which areas of the maritime industry AI has already changed and which are next in line. Dive into the future of maritime with Rory Proud and learn how to navigate the AI-driven transformation.
Key Takeaways from the Conversation
AI’s on the right course to maritime optimization: It enhances predictive destination tracking by integrating tracking data, weather events, and port congestion to provide accurate ETAs, addressing issues with AIS reliability. In maritime compliance, AI flags deceptive practices and predicts vessel activities during AIS outages, aiding legal navigation amid geopolitical shifts. Additionally, AI utilizes satellite imagery for insights into opaque markets and supports decarbonization, chartering, and email parsing for efficient maritime operations.
Data quality and resistance to change are the main obstacles: The maritime industry’s data can be big, messy, and easily manipulated, posing a significant challenge for effective AI implementation. Variations in vessel characteristics and ownership information are an additional layer of complication. The specifics of maritime industry doesn’t help either - being very traditional, it often demonstrates resistance against adopting new, especially automated, technologies.
AI and ML can solve major issues of the maritime industry: These technologies excel in processing partially unstructured data, making them highly suitable for maritime applications. AI can optimize routes, reduce costs, and enhance decision-making by processing vast amounts of data, providing a competitive edge for adopters.
AI will stay on an even keel: The future may bring significant investments in AI solutions for maritime logistics and supply chain management. As the shipping industry evolves and opens up to increasingly adopted AI solutions, its usage for linking disparate actors, normalizing data, and facilitating transparent communication may become more common.
Your company can benefit from it, too: Rory can show you the ropes, specifically in respect of which data may be most suitable to utilize for AI driven applications as well as a range of ready baked solutions, created and maintained by our partner network, such as predictive destinations, supply chain management and voyage optimizations.
Conversation with Rory Proud
Jarek Jarzębowski: Can you introduce yourself and your company?
Rory Proud: I am Rory Proud, a co-founder of Maritime Data. To give you a bit of background, I’ve worked in maritime intelligence for over ten years now. I spent most of my career at one of the market-leading providers, a company called Lloyd’s List Intelligence. I was lucky enough to be offered multiple opportunities to take on more responsibility and bigger roles.
While at Lloyd’s, I was running the global key accounts team before moving to New York City in 2020 to head up sales and account management for the Americas. I founded Maritime Data with another colleague from Lloyd’s in 2022. We’re a digital broker of data and analytics services for the maritime industry, now with a partner network of over 50 specialized providers.
For those who aren’t familiar with the size and scale of the shipping industry, there are around 250,000 unique vessels transmitting positions. We’ll touch more on that later. Along with the sheer number of ships, entire industries are built to serve this market and others explicitly linked to it. For example, ship brokers, ship suppliers, engine and equipment manufacturers, marine insurance, ship finance, fuel suppliers, and the list goes on.
Industries closely affected by shipping, like commodity trading, trade finance, integrated energy companies, even hedge funds and governments, need to understand the dynamics of shipping to operate effectively. You can imagine, with an industry of that size, a lot of people have discovered the potential to serve the wider maritime industry with digital services. Along with technology tailwinds like the proliferation of machine learning and AI and access to global talent, this has led to an explosion in the number and variety of solution providers in this space.
I’ve recently been appointed to the steering committee of a longstanding association called the International Maritime Statistics Forum. I gave a presentation on this topic at the last annual general meeting in London. The general idea of the presentation was that AI is lowering the barrier to building products through automated code generation, natural language processing, and generally acting as a force multiplier. This allows both incumbents to build faster and new entrants to create software solutions at a fraction of the cost and time, giving customers the option to build in-house should they choose. All of this is massively increasing the demand for high-quality proprietary data.
As a sidebar, as an industry, we need to make maritime data more accessible to help fuel innovation in the sector. Let me highlight three use cases where we see AI being used to great effect in the world of maritime data.
The first use case I’ll talk about is predictive destinations. To step back a bit, for readers not familiar with vessel tracking data, the real-time monitoring of ships around the world is predominantly done by a technology called AIS, which stands for Automatic Identification System. Every ship over 300 gross tons has one of these transponders on board and is tracked by a network of terrestrial, satellite and ship-borne receivers placed strategically around the world.
According to our evaluations of the leading AIS providers, there are roughly 250,000 unique vessels transmitting a signal at any given point in time. Roughly half of those report a position at least hourly when looking at high-frequency feeds.
The challenge with AIS data is that it isn’t always reliable. There are issues like signals getting jammed in congested areas - ports, trade lanes, and choke points. The technology is also open to manipulation, be it in the form of spoofing signals, positions, or simply entering the incorrect record when updating a ship’s ETA (estimated time of arrival). Sometimes these incorrect records are honest mistakes, such as entering the wrong port code. For example, instead of SG for Singapore, the captain might enter SH, or they simply don’t update it on time.
Shipping is a game based on one’s assessment of supply and demand. Broadly speaking, you want to have your ships positioned in areas of high demand and low supply to maximize freight rates, so disclosing your next move might not be in one’s best interest. This causes all sorts of logistical challenges to cargo owners, freight forwarders, ship supply businesses, and more. It was clearly a big enough problem for solution providers to create AI-powered predictive models that analyze not only tracking data but also weather events, congestion at ports, and other key indicators. These models constantly recalibrate and provide end users with a significantly more accurate ETA. I’ve seen figures quoted by customers suggesting an improvement of anywhere between 30 to 80% in accuracy versus AIS-derived ETA.
You’ve got ocean freight visibility providers like GNosisFreight,, Portcast, Beacon, and Vizion, all great examples of companies who’ve built and deployed solutions, as well as energy intelligence specialists like Kpler and Vortexa.
The second use case I’ll talk about is maritime compliance. In 2020, OFAC released a set of guidelines for companies operating in the maritime industry around the identification of deceptive shipping practices. Deceptive shipping practices are essentially a set of tactics used by bad actors to avoid sanctions while engaging in illicit activities. An example could be linking up with a vessel in the Persian Gulf and receiving a cargo of sanctioned oil before sailing back to Europe and selling it back into that marketplace.
The seven deceptive shipping practices highlighted are:
- disabling or manipulating AIS
- physically altering vessel identification
- falsifying cargo and vessel documentation
- ship-to-ship transfers, voyage irregularities
- false flags
- complex ownership
- complex management structures.
Solution providers created systems that can support predominantly banks, insurers, energy businesses, and governments, albeit the scope of their customer bases is expanding to other sectors. These systems not only flag when deceptive shipping practices are taking place but predict what those vessels are doing during periods of darkness - that is, when a ship turns off or manipulates its AIS position.
In this case, solution providers create descriptive insights to highlight when AIS is turned off or vessel identification characteristics change regularly, or there are voyage irregularities. Predictive insights are built on huge historical datasets and vessel trading patterns, essentially asking what is the probability that a ship within this timeframe could have engaged in sanctioned activity.
For example, could ship A, based on its max operating speed and the average turnaround time in port B, feasibly have collected a cargo and returned to the point it last transmitted a signal? You’re going from identifying potential deceptive shipping practices to predicting what these vessels are engaged in based on the amount of information we can collect and model, supported by AI and machine learning technologies.
Given the rapidly changing geopolitical climate and the events that have subsequently taken place since these guidelines were put in place, such as Russia-Ukraine and the issues in the Red Sea, sanctions are constantly changing. Solution providers have to innovate fast to create solutions that ensure their customers understand who their counterparties are and ultimately remain on the right side of the law. Companies like Windward and Lloyd’s List Intelligence wouldn’t be able to do this without the utilization of ML and AI.
As a sidebar, compliance products are unique in the sense that supply drives demand to a degree. Regulators are capped at the degree of transparency that’s realistic, so when organizations like Windward and Lloyd’s List Intelligence continue to improve their offerings, it changes what’s expected of businesses in the space. Given they’re able to develop and iterate products significantly faster with the utilization of tools and technologies like AI, that’s changing the very kind of regulations themselves, which is an interesting concept.
The third use case I’ll talk about is the use of satellite imagery to derive unique insights into otherwise opaque markets and situations. Utilizing satellite imagery bridges the disconnect between the wealth of information available via satellite data and those who can really use it.
I’ll give you three examples. Ursa Space, an organization based out of New York, is utilizing tens of millions of images to monitor oil inventories. They look at 22,000 tanks, six and a half million barrels in 123 different countries, providing critical data to traders, hedge funds, and governments on data that is otherwise unreported or unavailable.
Symax, another satellite intelligence company, collects daily images of the entire planet to detect maritime events and illicit activities. Similar to the use case described beforehand, rather than using AI to identify trading patterns and voyage irregularities, they use it for vessel identification. They analyze satellite imagery for attributes that can be linked back to a ship registry and ownership record to determine which vessels are in the respective images.
Tathya Earth, another partner of ours, has developed proprietary models using AI and ML based on synthetic aperture radar and infrared thermal imagery to monitor the production, mining, and inventory levels of industrial commodities like steel, iron ore, coal, copper, and zinc. Consequently, in response to trade conflicts, nations are increasingly gravitating towards siloed and protectionist economies. Essential economic indicators such as steel production data can be strategically manipulated, and there’s often a lag in reporting. Their algorithms determine the activity at a given site, which is then converted into indexes using different statistical models, providing near real-time and unbiased production, mining, and inventory data at a regional, company, or site level.
Other notable mentions of use cases impacted by the adoption of AI are decarbonization, chartering to a certain extent, and the advancements of email parsing, which is a whole other call. Companies like Ship Fix, and Signal Ocean are doing great work in that space. In summary, AI and ML are already prevalent in maritime across a group of critical use cases. AI will lower the barrier to build products, leading to even more solutions being built and more new players entering the space. For these players to produce high-quality, differentiated outputs, they’ll need quality, differentiated inputs, which will continue to increase the demand for maritime data.
Jarek Jarzębowski: From what you’re saying, there is a plethora of use cases in the maritime industry for AI and ML. With the experience you’ve gathered working with your partners and in your previous positions, have you seen or identified common challenges in implementing AI solutions in the maritime industry? Is there anything specific for the maritime industry that can be challenging for companies to innovate and implement AI solutions?
Rory Proud: I think one of the biggest challenges that maritime faces in implementing AI-powered solutions is the quality and structure of data that you’ve often got to build upon. I touched on some of the cons and inconsistencies of AIS, where it’s easily manipulated, voluminous (we’re talking about 400,000,000 records per day), and very messy. While it’s designed to be homogenous, many incorrect records make it through suppliers’ quality control checks, not providing the kind of structured databases that would be optimal for AI deployment.
Vessel characteristics are another example. These are foundational datasets that nearly every technology company needs to incorporate into AI-driven solutions. No organization has a single view on how big the world fleet is, which is a starting problem. They all have different characterizations of what a ship is, who the owner is, and what that owner’s role might be. Some call it a charterer, some a commercial manager, and others a third-party operator. One organization might describe a particular ship as a bulk carrier, while another might describe it as a general cargo vessel. The data itself is probably messier than ideal, and from the organizations we work with, there is a lot of work in data preparation and processing before you can effectively use these tools and technologies.
Jarek Jarzębowski: Even though it must be really messy and more difficult than if it were structured data, I think AI and ML are probably the right tools for that because traditional software needs really structured data.
AI and ML can deal with at least partially unstructured data, which is a big potential. Based on your experience, are there other areas that might be ripe for implementing AI in the future? Such that have not seen AI solutions yet, but will bring massive value to the industry?
Rory Proud: While I think about that, let me go back to your previous question about barriers to adoption. Shipping is a very traditional industry that has been around for a long time. As a result, you’ve got many long-standing businesses that have done things in a particular way for a long period of time.
One use case I mentioned was how AI is beginning to influence chartering. Chartering is essentially like loaning a ship, similar to renting a car. Companies are creating AI-driven email parsing tools because the industry still operates where everyone receives thousands of emails daily with either cargoes that need to be moved or ships in certain positions to move those cargoes. A broker’s job is to assess all the information from these emails and try to match supply with demand.
That’s a perfect problem for AI, which can pull out interesting information, structure it in a useful way, and augment those datasets to make them even more impactful. However, there is significant resistance to using automated technologies in longstanding traditional industries. Similar software has been around for some time, and I think AI has only recently turbocharged these solutions. But the general resistance to using automated technologies in traditional industries will always be a barrier for shipping.
Regarding the future use of AI, the five key areas I mentioned are where we’re seeing significant impact. However, there’s a lot of work going on in logistics and the supply chain to link up disparate actors. You’ve got multiple formats of data being transmitted, from PDFs to Excel spreadsheets to Word documents to emails. Consolidating and normalizing this data to facilitate transparent communication in the supply chain is a growing area, supported by significant investments.
Jarek Jarzębowski: Do you think that traditional resistance to new technologies in the maritime industry might lead to a disruption similar to other industries? Could new companies or startups disrupt the industry significantly?
Rory Proud: I don’t think it’s about new companies suddenly becoming shipowners and using technology to make decisions. It’s more about existing companies adopting these technologies to become significantly more competitive.
Look at fuel costs, which can be 30% of a ship’s operating expenses. Utilizing technology to optimize a vessel’s route and minimize costs translates to money on the bottom line. By crunching vast amounts of data from vessel tracking to seaborne cargo flows, companies can make more informed decisions about positioning their ships in areas of high demand and low supply. This level of decision-making, considering numerous moving parts and variables, is challenging for humans but ideal for AI. Companies adopting these technologies will start performing significantly better than those that don’t. The saying, “It’s not AI that takes people’s jobs, but people using AI that take people’s jobs,” applies to shipping as much as any other industry.
Jarek Jarzębowski: For companies with a more traditional approach wanting to use more data and modern technologies, do you have tips on how to start? What would be the first steps for them to adopt a more futuristic approach?
Rory Proud: It probably starts by defining the problem statement. Identify the most laborious or arduous parts of your job. For example, AI has been effectively deployed in trade finance and compliance, where compliance managers had to comb through disparate documentation to find necessary details. This process was labor-intensive and time-consuming. By scanning documents, parsing the information, and automating the checks, companies have significantly improved efficiency. Check out the work Trade Sun are doing for example.
So - start with a problem statement. Secondly, consider the numerous solution providers available. If you don’t have a defined problem or understand which part of your work can be optimized, seek help.
This is the problem we as a business aim to solve. We built Maritime Data to provide a single portal for exploratory work, defining problem statements, and matching them with appropriate solution providers. This allows companies to trial and test solutions to see if they optimize their business as hoped.
In summary, define your problem statement or reach out to solution providers or organizations like ours to navigate the technology landscape and explore use cases for automated technologies.
Jarek Jarzębowski: Great, I think this is the perfect place to ask where people can find you and learn more about Maritime Data.
Rory Proud: Of course, Maritime Data has just launched a new website (Maritimedata.ai) with extensive content on all our offerings, how to evaluate solutions, and the use cases they address. You can also follow us on social media, especially
LinkedIn, where we are very active.
Jarek Jarzębowski: Great, we’ll give the exact details. Thank you for coming prepared. I didn’t have to ask many questions because you were so well-prepared!
Rory Proud: Thanks a lot.
Rory Proud’s Background
With over a decade of experience in Maritime Intelligence,
Rory Proud excels in Business Development, Account Management, and Sales Leadership. His passion? Above all, helping the maritime industry leverage big data and analytics in a fast-evolving environment.
At Maritime Data, Rory co-leads a digital broker platform, connecting customers with top data and analytics solutions. He crafts and executes sales strategies, builds key stakeholder relationships, and oversees team operations and growth.
About Maritime Data
Maritime Data is revolutionizing how companies source, evaluate, and purchase data and analytics solutions within the maritime industry. By acting as an intermediary between buyers and suppliers, the company ensures that businesses can efficiently assess and procure the most relevant data products and services to meet their specific needs. This not only simplifies the procurement process but also enhances the accessibility of critical maritime intelligence.
Maritime Data’s mission is to empower companies to understand, access, and apply maritime data effectively. With increasing scrutiny on environmental practices and the dynamic geopolitical landscape, refines how companies engage with big data and analytics, supporting the development of solutions that drive compliance, efficiency, and sustainability.
Closing Thoughts
In the shipping sector, change is in motion - and as the development of AI progresses, we may see it become increasingly efficient in the coming decades. Data intelligence solutions, such as those provided by Maritime Data, not only save freight companies money and hassle but also enable them to implement stricter environmental measures.
The shipping industry facilitates over 80% of global trade, and the scale of vessel traffic makes optimization crucial in addressing climate change and congestion issues on most frequented routes. It’s all hands on deck as we watch which areas embrace new AI tools the quickest and what the consequences of this deep dive will be!
