Better Data for Shippers: How Nautilus Labs Is Helping Vessels Become More Efficient
In an industry as old as shipping, timeworn techniques and best practices can be hard to question — even as the marketplace shifts over time. But Nautilus Labs CEO Matt Heider says that owners and operators are, by and large, becoming more receptive than ever to the idea that they need to use data and analytics to better understand the vital sign as their vessels move around the world from port to port.
Heider caught up with SeaAhead recently to talk about how his company is helping shippers optimize speeds for efficiency, and why the investment landscape for bluetech startups is finally starting to take off.
Tell me a little about your background and how you got involved with Nautilus Labs.
Prior to Nautilus, I built the sales organization at a startup called Andela in New York. I joined in the very early days and helped to get it really launched, generating over $10 million in revenue a year — and now they’re doing amazing things. A pretty inspiring company. Prior to that, I had worked for a number of years at IBM and sold analytics solutions, Watson-based products into a number of different verticals. So I had seen a lot of different forms of machine-learning-based decision support applied to a variety of industry challenges in the telco and consumer retail space.
Nautilus was kind of fusing those two things together — a background as leader in an early-stage startup and getting a business off the ground, and also understanding the application of machine learning to decision support in various industries.
What we all coalesced around here at Nautilus in its early days — and really the core thing that got everybody engaged and interested in the company — was when we began to understand how little data was used in decision making inside [shipping] companies. Mainly, we focus on ship owners and operators.
We came upon their core fundamental fallacy of relying upon noon reports to inform all forms of analysis that occur in all sorts of business decisions. We realized that there needed to be a better system of record, and a better source of information upon which to base this range of decisions, as well as a tool to help make that data usable for all the different users that persist inside these organizations.
That’s really where we started, working on this fundamental challenge of how and where to source better data. Then we developed a platform that could integrate a range of data sources, building in basic analytics that could provide decent operational insight to folks who hadn’t had it before. Now we’re increasingly working on advanced forms of decision support that leverage machine learning and can create a range of different optimizations for a vessel or a voyage across entire fleets.
How did you come to see the shipping industry as a focus?
I had come to New York to do the startup thing and luckily had been successful in my first startup, so I had some confidence that I wanted to go do it again. When I met the co-founders, I found out about this challenge in the industry around data — and it was a really interesting problem to solve from a technological perspective and also from a market perspective. How do you grow and scale a business in an industry that is so global?
Shipping touches us all, and no one thinks about it. When I first got acquainted with some of these sort of foundational issues in the industry it was not because I had spent a lot of time thinking about maritime shipping. It hadn’t really crossed my mind. But you begin to realize how it touches everyone and really everything that you’ve ever cared about or will care about, and every person you ever will care about or have cared about — it impacts us all. The fact that it’s not paid attention to and people don’t spend a lot of thought on it is something that has to get corrected — particularly when you think about the environmental impact shipping has on the world.
The shipping industry is obviously a very old and well-established industry. I would imagine there would be a lot of resistance to change. In general, how has the industry reacted to your product?
That’s a really good question. What we find overall is that the industry is open to adopting new technology now, and the conversation that we have today is different than the conversation that we had a year-and-a-half ago. A big part of that is IMO 2020 and the pressure that’s on owners to find new and better ways to do things.
We could have discussion about scrubbers and whether or not scrubbers are a wise investment, and what the return is on them and all of that. Fundamentally, scrubbers don’t get to the root of the problem, which is twofold. One, the industry consumes a lot of very pollutant fuel. And two, the price for fuel is going to go up remarkably in a couple years. I think owners are realizing they have to tap into something new to solve it, and are open to considering new technologies.
There is a certain amount of education required about things like machine learning and high-frequency data to go a little bit deeper into how or why it’s different and better than the methodologies that have been used in the past. That is an ongoing conversation that you have to have with potential partners and clients about it. But, overall, I think just the changing market dynamics for fuel are compelling a lot of people to consider new options.
In general, the thing that we’ve heard repeatedly — even before we ever had a client — was that folks in the industry believe that change was going to come from outside the industry. So there’s been an expectation for some time that it would be players from outside who would come in and bring some of this innovation and change to the space.
Can we dig a little bit deeper into the machine learning mechanisms and AI? What kind of data are you surfacing here and how is it being used?
We source data from a few different places. One is what we would consider to be vessel IOT data — “internet of things” or sometimes called “industrial internet of things.” That’s all the sensors that can persist on a ship from its GPS sensor to fuel flow meter to the shaft torque meter on the engine, draft sensors, navigational signals, all of this type of data is sensors on a ship and can be collected, aggregated, and sent off the ship for analysis.
It starts with that foundational high-frequency dataset. We combine it with some other sources of information that we tap into ourselves, a primary one being all the data we get about the sea and weather states around the ship — wave heights, swell, wind direction, air temperature, water temperature. These types of data points are really helpful for contextualizing ship performance. One of the hardest things about performance analysis is actually understanding true performance inside the context of weather. Then we’re also pointing AIS data to do some sort of corroboration and other analysis that ties back to high-frequency analysis.
Finally, we have data that’s supplied by the clients themselves — things like the commercial terms from the vessel, market rates, market prices. What our platform does is fuse this all together in one place. Machine learning right now largely plays into two different algorithmic things that we’re doing. One is the normalization of performance — taking how a ship has been performing over the last three, six months, a year; stripping out the effects of weather, and predicting what the expected performance will be in a given draft condition, in a given weather condition; and then applying that to a set of optimizations that we can run. One of the primary ones is predicting the optimal RPM speed for a ship to run at.
Specifically for fuel economy?
It often comes back to fuel economy. It depends on the segment that a ship is operating in and the type of charter contract it’s on, but in general, most of the time, the most basic and most requested optimization is just: “If I can arrive in a given timeframe or at a given time what is the slowest speed or RPM I can run at I order to get there reasonably on time?” The optimal RPM to run at constantly over the course of the voyage to get to a given destination at a given time.
Typically that ends up being a fuel cost minimization calculation, which is the most efficient thing for all parties involved. We’ve built a model that pulls in other market dynamics and market factors. So if you’re attempting to arrive in a port to discharge and you know what forward market rates are in that port and you’re on voyage contracts, you may elect to run a little bit faster to get there more quickly so you could discharge and pick up a new load and remarket your vessel. It improves operational fuel efficiency and enables shoreside teams to more accurately vessel performance, and ultimately command a better rate in the market.
The bluetech industry is still pretty nascent, and still doesn’t really get the kind of VC focus that other sectors get. Did that have any effect on Nautilus’s development and growth? Tell me a little bit about the process of raising capital and prototyping to then getting the product out there.
I think bluetech has received insufficient attention from the tech industry more broadly and I think that’s what has inhibited investment to a certain extent. VCs go where they can find and capture positive returns, and there’s not enough smart, thoughtful, well-intentioned focus to bring new technologies to maritime.
So I think as people come to the table [with big ideas], they’re in turn seeing more and more VCs beginning to pay attention to maritime. More folks are getting engaged with the problem sets that exist in the industry, and in turn are finding ways to solve problems that add value that attract the capital interest.
As we think about bringing in more capital in the future, I think we’re engaged with the right people. Like any other startup in New York, ultimately the question [from investors] is whether there is a potential to create a large impact. Are you solving a real problem and have you demonstrated that you’ve found a method that solves the problem? I think they’re kind of agnostic as to whether that’s bluetech or dog-walking or free movies. It’s more about a logical revenue model associated with the business that you’re building.
Did you have any advice for other entrepreneurs looking to get into bluetech? What are some of the major hurdles?
If I had to choose one thing, the word I would use is partner. From our very early days we have done an extremely good job partnering with our clients first and foremost. It’s about really treating our relationship with our clients humbly and going to them and inquiring as to what they want and need, and not taking a preordained approach to telling them what was the right solution. Instead, it’s working with them to cover all bases and find the right solution for them. I think if anything could have been make or break for us in very early day it probably would have been having a mentality that “We’re right, you’re wrong. We’re from Silicon Valley, you’re not. This is what you need. This is not what you need.”
In line with that, we also very early on had some strategic partners who helped us make good decisions. When you’re developing a technology like ours, the first ship that you get on and the first fleet that you engage with is always the hardest challenge. So it comes back to that idea of partnership, which is figure out and engage an advisor who is a strategic fit. They help you strategically get into your first engagement where you can identify the problems, generate a solution and figure out how you can bring that to other potential clients down the road.
Going along with that, just know whom you're helping and why. Since early on, we’ve always been really clear in the company that we help owner-operators solve fleet performance in an optimized way for their vessels and their organizations. That’s been a running theme for us even from before we actually had a software platform to show anybody.
One thing that I’ve seen from companies in a variety of different technology spaces — and definitely in ours — is a desire to help everyone with everything. I really think you can’t be everything to everyone, particularly when you’re a young company with strained resources. So really understand who you want to help, who you can help, and be very clear about that to them and to everybody else so that people know what you can do and how you can do it. I think that desire to do too much, to be spread too thin, to try to solve all of the world’s challenges at once, can often be the thing that brings a company down.
David Hirschman is SeaAhead’s VP of content. This interview has been edited for length and clarity.