In this interview, Antonia Saratsopoulou speaks with Adam Dennett, CEO of SpecTec and Chair of the Society of Maritime Industries’ Digital Technology Group in the UK. They discuss how AI is reshaping maritime operations, the importance of data standardisation and interoperability, and why shipping companies must focus on solving operational challenges rather than simply adopting technology trends.
The conversation also explores predictive maintenance, human-first AI integration, and the growing role of digital platforms in improving fleet efficiency, reliability, and decision-making across the maritime industry.
- Across the maritime industry, there is enormous excitement around AI right now. In your view, what are the biggest misconceptions companies still have when it comes to implementing AI within shipping operations?
There is a real unjustified reason around that excitement for AI in maritime. I’ve been two years at SpecTec now, but prior to that I’ve spent my career in technology. For those that have been around a little bit longer as well, they will have experienced the internet coming in, wireless connectivity, Bitcoin.
But ultimately what we see, AI is probably the largest technology shift that any of us have ever experienced in our lives, which for me can be scary, but ultimately really exciting as well.
We’ve worked closely with a number of companies at various stages of their digital journey. I’ve also seen a handful of those misconceptions. I think that you mentioned that consistently can lead to slow progress and disappointment, and I think one of those is where people will treat AI as the starting point.
So, there’s a lot of companies saying we need to do AI, but they don’t really know why and what they’re trying to do.
A challenge with applying technology like AI can be around the data that can be obviously deeply siloed in shipping, which can be historic due to connectivity.
When systems don’t talk to each other, if you don’t have that foundation of standardized and interoperable data, the limited potential you can get from just applying AI solutions will be limited ultimately.
A step that people need to look at first, before they just simply jump into AI solutions, is understanding their data frameworks, where they have data, how systems talk, and then the other real misconception is probably the volume of
activity equal’s progress.
The whole market is very busy looking at these solutions, but there were some interesting stats from a sort of the Massachusetts Institute of Technology. They stated that 95% of Gen AI pilots actually failed to deliver impact. And I think Gartner was producing similar stats, basically saying that 40% of AI projects, agentic AI projects will fail by 2027.
That’s not because AI doesn’t work. AI has huge potential, but ultimately people are applying the wrong technology to apply the wrong shape of technology to the wrong problem, or trying to do it alone and simply working with organizations that don’t really understand the industry or their ways of working as well.
- The maritime industry includes operators with very different fleet sizes, operational requirements and levels of digital maturity. In your view, how should companies evaluate which AI tools genuinely fit their specific operational needs?
When you’re looking at that, the thing I would always stress, and it has been common since I’ve been working in tech, is that always start with the problem, not the technology.
So, most companies that get the most out of AI, they start with a specific operational pain point, and then they work backwards towards them finding goal, building the right solution. Critically, we’d always advise, “don’t try and do this alone as well”.
So, partner with solutions providers, share your problems, honestly. And the best conversations that I have with customers are when we start with, “here is what is broken in our operation”, or “this is the problem we’re trying to solve”. They don’t come to you saying, “oh, here’s the AI feature I want to buy”.
Because I think when you look at those problems, some problems just need automation. They might be a rule-based, repeatable process that doesn’t need AI, just needs automation. Some might need surfaced insight where AI can do that heavy lifting on the data and the human can still make the decision.
Some can then apply that true agentic capability. And that’s where the system can take action across multiple systems on behalf of an operator. And then technologies are also about changing how people interact with systems entirely as well.
Ultimately, a port captain, a superintendent can ask a question and get an answer in their own words. So, when they’re looking at what tools they evaluate, I think really we must focus on that problem. Sometimes that problem and the most near term opportunities aren’t necessarily the most glamorous ones or feel exciting.
They’re generally where you’ve got high volume, resource intensive and like low value tasks that really tie up your experienced team and qualified people. Seafarers, shore professionals, doing work that they shouldn’t be doing.
So, the big area we found recently is around like manual invoice matching, reconciling spare part records, that carry different names across vessels, chasing purchase orders, rekeying data between systems.
So, it’s those areas where I would like to deploy AI first, because every hour that you get back, can then be given back to those really qualified and professional individuals to do the work where they add real value.
That’s an area where we’ve built Amos Procurement Smart, which is really going to help attack that challenges, operators have. And then predictive maintenance, is the other real clear opportunity because, given the fleet is at its average oldest age, that it’s been in 40 years, older assets, regardless of their segment or fleet that they operate, they generally have higher insurance claims, their prone to more safety risks, small parts failures.
So, being able to shift from that time to condition based maintenance, is where really then AI can really help analyze that data across thousands, if not millions of data points, and really guide the sort of the operators in the right way. So in both cases, it’s that partnership matters as much as technology.
- SpecTec operates heavily in the area of asset and maintenance management through the AMOS platform. How do you see AI reshaping predictive maintenance over the next five years, particularly for vessel operators looking to reduce downtime and improve reliability?
The way AI is really going to help is around the predictive maintenance algorithms, so the data points and how it makes the decision and the understanding that it has. You’ll also see that assisted like digital twin modelling, which we talked around a lot as well, and then failure mode and effective and effects analysis, which it really helps for that proactive risk mitigation.
So, what the AI tooling can then ultimately do, is detect anomalies, easy for me to say, anomalies in engine performance forecast component where it can also then allow procurement systems to automatically initiate replacement orders, and then let
inventory platforms, track usage trends, optimize stock levels.
So for me, the value is not just in maintenance cost, which it can have a huge benefit in, but it’s that schedule integrity because unplanned downtime in any container service is a missed birth window, that missed transshipment and contractual penalties.
So, AI’s contribution over the next five years, and I think we’ll see it a lot sooner than five years. I think the tech landscape is moving at such a pace, three to five years is becoming what’s happening in the next 18 months. But the key benefit will be making schedules more reliable, and not just engines more durable.
That sort of that same standardized data foundation, is needed and applied. So as models become more sophisticated and process more operational data, the recommendations will get sharper with time and the applications then can spread further across the business.
Where this can get, I think, really neat is when data sharing is enabled as well between operators, which is a barrier that we would need to cross.
But for example, if an operator can benchmark their same components across sister ships that they may own, but also other operators may own, worldwide, you can then have that performance pattern data and sit across like fleet wide evidence bases and sort of worldwide evidence bases rather than just your operation.
So I think for those sorts of mid-tier operators as well, where they might have 10, 15 vessels, being able to compare your parts across
1000 vessels, 3000 vessels will enable you to make a lot more better decisions as well.
So, you can make better maintenance calls, earlier site of like systemic issues and really put you in a stronger position with OEMs as well.
- AMOS-X represents a major step in SpecTec’s digital evolution. What were the key industry challenges or operational gaps you aimed to address through this next-generation platform?
We didn’t introduce just new technology for its own sake. The aim is really building something rooted in that operational reality across maintenance, inventory, procurement and QSHE as well.
When a component might be flagged for replacement, the maintenance team know, but procurement don’t. And say the inventory then may not reflect what’s actually on board. So the unified AMOS platform really brings all of that together into a single interoperable environment.
So that information flows automatically and then the right action follows the right insight and really reduces manual human intervention, which was just tying up people unnecessary.
We’ve also really designed it for the operational reality, our customers working end to end. Shipping has probably spent 30 years building software for either the back office, specifically, but we’ve really built it for both back office, the technical team, procurement, finance functions, the bridge, all in one platform.
So, it’s a human first and device agnostic in terms of how you can use it. A superintendent can be at a desk, a buyer can be on a tablet, an engineer can be on the phone at sea, but they all get an experience that’s optimized for where they are and what they’re trying to do.
Because we know when applying any or introducing any new tech platforms, adoption is where most platforms fail. So that’s why that usability element of the platform is really non-negotiable, and that interface matters just as much as the algorithms and the functionality available.
So,is multilingual by default to reflect the reality of a global workforce. It’s responsive on whatever device is in front of them and it’s really designed to guide work in the moment rather than just capture it after the fact. The shift is from that system of record to a system of understanding.
The software has more context and validates the data it holds, and then that can really help those shore and those ship teams make the next decision a lot easier.
- One challenge frequently discussed in maritime digitalisation is fragmented and inconsistent data. How critical is data standardisation before companies can truly unlock the value of AI and advanced analytics?
Critical and important. It’s so key because if you have data that’s not interoperable or it’s in separate places, then you’re ultimately asking sort of AI to be able to interpret data points in silos. It doesn’t have the full picture.
Therefore it’s like you making a decision on one piece of data rather than free. So, you could probably make a better decision. If you simply apply AI to just procurement in isolation, it’s probably not going to add as much value as it could if it had exposure to all of the other areas.
And we did some recent research, there was 20 fleet operators that represented more than 3000 vessels. And we dug into some of the procurement processes and they were still managing thousands of RFQs, purchase orders, invoices, spare part requests, all manually.
And you sort of discovered that the same component can carry 12 to 20 different names across the fleet.
So, it’s the same thing, but people just refer to it differently, and invoice matching failure rates were exceeding like 60%. That was just one example. So it’s not standardized in one system, never mind the silo barriers that we have. I would really say to sort of any owner that watching, this is that
solving that data problem. And shipping really talked about cleaning up its data for the last 15 years, and probably never has really found the business case to do it, but AI is that business case and can really help as well.
So, it’s not just about the AI tool, but you can use AI to help you clean up that data problem as well. And I think once you get that foundation in place, you
standardize your components, unify your maintenance of procurement records, get that one version of truth across your fleet, that will then enable you to unlock more of that capability that comes to life. Once you give AI that sort of guide and access to what it needs, AI can start becoming more of a talent within your organization than just a simple add-on tool. It can start to drive decisions for you as well.
- Smaller shipowners and operators often worry that AI innovation is designed primarily for large global fleets. Do you believe scalable AI solutions are becoming more accessible for smaller operators as well?
Some of those bigger operators can have more resilience to obviously fluctuating market demands, price changes, etc.
But margins are tight across shipping and across the board, so it can really hurt those smaller operators more than ever.
Generally, you’ll find those small ship owners operating across multiple regulated jurisdictions, they’re doing diverse trade routes, and they don’t always have those in-house capabilities as well that a bigger owner might have. We sort of say that segment of the market within 10 to 20 vessel range, where the majority of the industry sits.
They really need solutions that are built for their needs, not scaled down versions of enterprise tools. And I think we are seeing with AI is very sort of becoming more price competitive in the market as well, which is great.
There are advantages, generally able to deploy tools much faster. So, we’re seeing more of an appetite from that segment to adopt digital solutions to
engage in discovery to co-create and build with us.
We’re really proactively supporting them, as much as we can by sort of lowering barriers of adoption and offering solutions that fit their needs.
So, we’ve recently sort of changed our packaging structure, in terms of how we offer our services and we’ve created sort of three tiers of packaging.
We have a gateway package, horizon package and then enterprise. So, we can give access to those sort of smaller tier users to the same level of functionality, to the same benefits of what the system can offer, but suited for them.
So, I think we’re seeing more of that come to light. And with web technologies and more SAS models, it’s easier to deploy technology than it might have previously been for that smaller operator.
- As Chair of the Society of Maritime Industries’ Digital Technology Group, you have visibility across the wider maritime technology ecosystem. Which areas of maritime operations do you believe are currently seeing the most practical and measurable benefits from AI adoption?
We’re also part of a global group of vertical companies. I’ve got the maritime experience. I can also dig into healthcare, local government, and logistics. And I think what I’ve really seen is that there’s areas where maritime is ahead. There’s also areas where were behind. One of the key enabling factors is those industries that have better data standards across the board.
Companies will really start small, they learn in public and they ultimately scale what works, but I think the ones who really struggle or don’t see those benefits, are ones that really try and get their way out of a problem.
Some of the key things we’ve seen is where the data foundations really are mature enough to support real models. AI assisted pilotage and bridge decision support is one of those.
Sort of port call optimizations, birth scheduling, voyage planning, that’s definitely another.
You’ve also got real fuel emissions, sort of scheduling integrity, the numbers behind it, and also insurance claims, like analytics. So quantifiable loss reduction starting to show up in sort of operator and underwriter conversations.
And then for us, the biggest benefits we see is, the world that we’re in is really that predictive maintenance and managing that full asset life cycle, which really sits firmly within the group.
With the age profile of the fleet, that’s been the area that I’ve definitely seen the most benefit that our customers have seen and experienced and operators through the board.
What really these have in common is not necessarily it’s the algorithm, but it’s the underlying data. Because AI is obviously being leveraged and more tech, is cyber resilience alongside the use of more digital tools, that is really important and we need to make sure that they really advance together, because one without the other can create vulnerabilities.
So that’s really a sort of change management challenge, and it really belongs at the board level, not just your IT department.
It’s just really making sure you are leaning into the adoption, the innovation, but just the exposure risk is probably a key part to make sure you’ve got in the back of your mind.
- There is increasing discussion around connecting maintenance systems, voyage data and operational analytics into one ecosystem. How important is interoperability between platforms in building the next generation of smart fleet operations?
That interoperability is vital across the platforms because we’re really experiencing significant regulatory operational change at the same time.
So, we know complaints frameworks are raising the bar on documentation, process consistency. You’ve got ESG and emission obligations under Fuel EU, UETS,
and they’re really creating that pressure on data quality and traceability across fleets.
As organisations grow in size and complexity, that cost and the impact of having systems that are fragmented and not interoperable, really compound quickly.
Because if you’re an organisation, you need to run as one organization and therefore
really the platforms that you rely on need, to behave in the same way.
So, they need to speak in one consistent voice, in a language across both asset, voyage, procurement, finance and crew.
That opportunity is really to get that interoperability right, which will then give you that transparency across the business.
And then if you’re using shared data, you can then drive ultimately better and faster decisions on the things that actually matter across sort of your safety, reliability, scheduling, integrity and cost.
So, the wider technology world has really already settled this argument. So, we can see that the platforms that win right now, are the ones that open doors to their
applications via APIs, and not closing it. Outside of sort of Maritime, you’ve got Anthropic and Microsoft, and they’re leading sort of cloud and AI providers. They’re deliberately making it interoperable and deliberately open because the value really sits in, the value to the customer, sits in what they can connect to, not what they’re locked into.
So, maritime vendors that try and build walls around their data, or they pull up the drawbridge, they’re going to find themselves on the wrong side of that shift quickly. So, customers won’t accept it, nor will they buy it. That connectivity issue is where the next wave of value sits. So, compliance reporting is forcing operators to really bring asset voyage and fuel data into one place. Then when you bring all of that data into one place in one language, you have platforms talking to each other. That’s where the magic can happen, in terms of the value that those operators could unlock.
- Shipping has traditionally been cautious when adopting new technologies. Do you think the industry’s attitude toward AI is now changing, especially as operators face pressure around efficiency, costs and compliance?
We’ve definitely seen the attitude changing, and the uptake of technologies and people willing to try stuff.
We’ve probably seen it all. So, when you see the value those tools bring to you personally, then there can be that more willingness to use them in business.
Sometimes we’ve seen technology is just a business problem, but I think we’re more open to that.
So really that hype and that awareness is real now. Industrial applications of that technology are really that natural progression. And as you say, you know, tightening cost control, margins, volatility in the market, means that you’ve got to be using the best technology to solve your problems.
So, you get that competitive edge and you can have that competitive advantage. It’s still fair to say that shipping is a risk averse industry, and the stakes are high for ship owners and operators, because you can’t afford to put lives and assets at risk due to an AI driven error or decision.
AI’s role in maritime is to support and complement. And we often talk about augmenting human decision making, not replacing it.
So, always keeping that human in the loop. All those technology providers are able to articulate that to the operators, they can see it. I think that gives them more confidence to adopt. I think when they just say, “oh, the AI has made that decision,
don’t worry about it”. that can be a bit risky, but we’re definitely seeing more of an uptake in terms of adoption of new attack.
- AI is often discussed as a tool for automation, but you have also emphasised the importance of human-first technology. How should maritime companies balance automation with the experience and decision-making of onboard crews and shore teams?
The goal when we look at it, is really using AI technology and tools to really give experienced crews and sure team sharper and like faster insights, so they can make decisions that are better informed. It’s not to take it out of the human hands entirely.
Technology that removes people from the loop too quickly will create new risks in an industry, especially where conditions are complex, they’re dynamic, and they’re often unpredictable in the maritime world.
So, technology crews do not trust, and they’re not using that tool,
day in, day out, and they don’t understand the framework.
So, for me, I think automation really earns its place, when it’s brought in incrementally
So, it’s brought in over time. It’s not full automation of a workflow, but you start to do it step by step. And you really start by taking away those high volume, repetitive tasks, which starts to give people trust in what the system is doing.
Then the human expertise is really reserved for them where it genuinely matters most, where they need to make a decision or approve a decision that the AI might have made.
And that same principle applies to the relationship between operators and their technology partners. Technology that is bought to a brief never quite fit. It’s about making sure you build that in partnership. And so, sharing your real pain points and the providers, it’s like how you build it in a human way.
That’s what actually transforms that process. So, we treat our customers as collaborators and not as buyers.
You’ve really got to show and prove the value and bring the people in, bring the operational teams in, the co-creation of the software, as well as the utilization of it.
And if the technology is as good as you say it is, you’ll generally find that people will adopt it and use it. So, it’s just making sure we always have that balance and we never lose sight of the value that the human mind and human judgment can really bring to the decisions, that seafarers and the crew worldwide are making every day.
- Looking ahead, what do you believe will separate the maritime organisations that successfully scale AI integration from those that struggle to move beyond experimentation and pilot projects?
It comes back to my point at the start and it’s really important that you start with the problem and not the technology.
So what problem are you trying to solve? And also, another key part is making sure as a business, that you run them like business transformations, not simply an IT project.
And so we often see that barrier in larger operators, where you have operational teams and IT teams, and they don’t necessarily align.
Companies that will be able to scale will probably do four things. I think they’ll start with that clear operational problem, not just a tool. And they’ll co-create and work collaboratively with the the solutions provider, sharing what’s broken and what solutions they’d like to see.
I think it matches the right shape of technology to the problem. So, is it automation that’s needed? Is it AI? Is it a natural language interface? Is it a mix of those? And then the other part is really treating decisions as two way doors as well.
A lot of people make decisions on IT in fear that they can never revert back. So, you might be able to just try and see if it proves and then always know that you should be able to reverse those decisions, and build that option in at the start. Companies that struggle, are ones that really want complete certainty before they commit.
Because, trying to get to that everything is 100% accurate is probably going to be challenging. So, you have to try with some degree of risk and tolerance as long as you understand what that risk is. But the other ones that may struggle are ones that don’t look at it as a cultural shift and it’s a leadership shift.
People are scared by this. We’ve been applying AI tools in our work and people can be fearful. So, you know, no one’s an expert in this, I would say. So, it’s been able to just share with humility, to have an open conversation, and willing to fail as well sometimes, like that’s really important.
That’s just really key on being able to take projects, pilot willingly, not running a culture of fear and being able to sort of move that forward.
But I think really the final piece is that AI can allow you to probably move it
a faster pace, be able to deploy more technology, but we can never replace what good tech does and usability is a key thing.
So, you can have the best product in the world, but if it’s not usable, people aren’t going to use it and it won’t get used. Making sure that solutions are well designed, they’re intuitive on the device that the user has in front of them.
They’re quick, they’re available in the language that they want to work in and they’re really useful. So, people will pull good technology into their workflow and will always find a route around bad technology when it’s applied.
So, no matter how clever that model is underneath it, If it isn’t easy to use, people will find a way not to use it.
Whereas if the software makes a difference, people will be willing to use it. And that’s really important. So, the vendors really will win in the next five years and be the ones that treat that ease of use, as a first of engineering discipline and not just sort of a UX polish at the end fundamentally.
