Full Steam Ahead: OrbitMI’s AI Maritime Routing Engine

Build or buy? It’s a familiar question in the tech world. For OrbitMI the answer would have a huge impact on its operations and the future of maritime supply chain management technology.

The company already had a third-party vendor supplying the maritime routing data behind its SaaS platform. But OrbitMI’s search for a better-performing replacement was fruitless; among their limited vendor choices, they already had the best option money could buy. Maritime routing vendors were simply not prepared for the complexities of today’s world. “They were limited by the creativity and the tools of the past. We didn’t find anything that actually met our needs today,” said OrbitMI CEO Ali Riaz.

So Riaz and his team decided to innovate and build something new that would meet the needs of today, and tomorrow, too. They found an innovation partner in Neo4j.

By the numbers: OrbitMI

  • Maritime route planning speed: <1 second
  • Route planning productivity: 60% increase
  • Platform: Neo4j Aura Enterprise on AWS and Neo4j Graph Data Science

With a focus on scale, agility, and performance, OrbitMI’s ambition was to create a software solution for maritime supply chain management that uses artificial intelligence, and integrates current and historical vessel positions, and multiple data feeds and APIs.

“We created something that did not exist in the market,” Riaz said of the firm’s custom AI maritime routing technology. “The whole reason we reached out to Neo4j was because we wanted to do some hardcore innovation.”

What Neo4j Does That Others Cannot

The OrbitMI use case was to create a reliable solution to serve optimized routes in a fraction of a second.

A graph data science framework was the clear best choice to solve the problem, but the scale of the data OrbitMI required that the solution be optimized to handle enormous volumes. Spatial data support was also key; it was a must-have, non-negotiable for OrbitMI.

To create a scalable graph solution that processes millions of nodes and edges, OrbitMI identified five criteria necessary for building its differentiated maritime routing solution:

  1. The system needed to be capable of processing huge volumes of data
  2. Data storage and retrieval had to be reliable, execute in real-time, and work at scale
  3. Support for spatial — in addition to linear and tabular — datasets
  4. A library of pathfinding algorithms
  5. A development partner willing to experiment and innovate

OrbitMI’s use case did indeed require much iteration and innovation, but the efforts and results were remarkable, and impressive enough that the firm earned Neo4j’s 2­021 Graphie Award for Graph Ecosystem – Supply Chain Excellence.

“Most providers fell off after No. 2. None were left by the time we got through No.5; Neo4j was the clear choice,” said David Levy, CMO, in a blog describing OrbitMI’s Graphie Award win.

A Maritime Supply Chain Graph of Leviathan Proportions

There are multiple, interconnected layers to the datasets used in supply chain management. Information on cargo, vessels, ports of call, regulations, weather, and location, just to name some. And like any industry, supply chain management is only getting more complex.

Whenever new data gets added to the graph it must get connected to data that’s already there, building a network of connected nodes.

“Every time there’s a new dataset that goes into our system, there’s an exponential scale required on that. It’s not just plus one. It’s plus one, plus all the other layers,” said Raiz. “So, we believe that the graph is important and it is going to be increasingly important to have the best graph technology to scale with.”

Running a Tight Ship (or 10,000)

OrbitMI uses Neo4j Graph Data Science to deploy the A* pathfinding algorithm to run its maritime route finding application.

Whether it’s hyper-optimizing one vessel’s route to make the journey as efficient as possible, or plotting the routes for tens of thousands of vessels through a single port and estimating when each vessel will arrive based on real-time data – it’s all powered by the graph database and algorithms provided by Neo4j.

“So far the results are great,” said Slavisa Djokic, VP of Engineering at OrbitMI. “And we want to push those results even more, since we are increasing the number of vessels routed, we are increasing the number of calls to our API and how those calls are being served by our API.”

OrbitMI’s innovative configuration of Neo4j puts them at the leading edge of technology for the maritime routing marketplace. While there are other firms in the space that can route a single vessel, Djokic said those routing tools and maps are not as precise as OrbitMI’s product. When it comes to projecting routes for multiple vessels heading to a specific port, OrbitMI is in a league of its own thanks to the competitive advantage it gets by leveraging Neo4j Graph Data Science. “To be honest, in our years on the market, we haven’t seen a solution that had this kind of use case,” he said. “So there is nothing to compare with, at least that I know of.”

A True Innovation Partner

OrbitMI knew that it didn’t want to go on its innovation journey alone. They wanted a partner who was willing to take the journey with them and discover along the way.

“We did not know exactly what we needed when we started the process; we had to iterate to get to that,” Riaz said. “Neo4j was willing to actually be part of the innovation process, not just dump some code over to us and wish just good luck. That was not enough for us because when you’re innovating, you don’t know exactly where you’re going. You just know you’re going to have a lot of iterations. And that’s what we received from the Neo4j team.”

Creating a Net New Capability

Now, with the size and reliability of the OrbitMI graph, the firm is able to get what it needs out of its data, and more. There are now enough data points in the graph for predictive analytics, such as computing where vessels will be in the future. And because they have up-to-date, real-time route data flowing through its powerful graph database, OrbitMI can do even more with its data than they initially imagined.

This was net new functionality for the industry. “Because that feature was available on the market, nobody knew that they wanted it. As soon as we presented it, it was like a mobile phone,” said Djokic. Customers were amazed to be able to see when their fleet would arrive in port. It was a process they were doing manually for each vessel. “It was groundbreaking for them,” he said.

“We are now capable of tripling the size of our workloads,” said Djokic. “All of that is because of this engine that is our routing application.”

Riaz echoed the sentiment, adding, “There’s an incredible turning point when you can do some things in a completely different way.”