Determining the optimal path for these types of shipments is extremely difficult. Most start at an arbitrary place—a lab, hospital, warehouse—versus an airport.
This interview with Ryan Rusnak gives us a walkthrough of AI and machine transforming shipping business.
1. Tell us about your role at Airspace?
I am co-founder and CTO at Airspace, a global delivery network focused on leveraging technology and people to make shipping faster, safer, and more transparent.
2. Can you elaborate on your career in the industry?
Before co-founding Airspace, I wrote software solutions for the Federal Government. I worked on an application that creates the federal budget, The White House iOS application under The Barack Obama administration, and early machine learning work with the National Institutes of Health. Throughout my career, I have also built robots and other tangible computing projects. Some of these have been featured in Popular Science, the Discovery Channel, NBC, BBC, WIRED, and Gizmodo.
3. Can you tell us about your journey into the logistics industry?
After writing software in the federal space, I was looking for the absolute hardest thing that I could build next that would be the best use of my skills and have the biggest impact on the world.
When I met Nick Bulcao (the CEO of Airspace), it really checked all of the boxes. The logistics industry problems we are trying to solve at Airspace have the complexity of continually tackling a NP hard, or needle-in-the-haystack-type of, computational problem, with literally hundreds of millions of suboptimal routes and trying to find that right one over and over again. And there is also the ability to make a huge impact in shipping some of the world’s most precious cargo—such as vaccines, critical lab results, and organs for transplant. It feels good getting up every morning knowing that the code we write has a profound effect on the world.
4. What are some of the challenges with shipping time-critical items?
Today we expect shipments for most everything to arrive safely and on time. But some things are more time-critical than others, for instance automotive manufacturing machinery, costing millions in downtime; lab specimens; and human organs. These shipments absolutely have to arrive securely and on time.
Determining the optimal path for these types of shipments is extremely difficult. Most start at an arbitrary place—a lab, hospital, warehouse—versus an airport. And then there are numerous options to take the package to its final destination—over 180 million ways to go just from San Francisco to Boston—as one example. In addition to choosing the best route, factors like traffic, weather, the driver’s knowledge of the route, and real-time incidents also need to be considered as events unfold, making it a highly-complex and ever-evolving equation.
On top of that, events can change. Consider the impact of inclement weather and flights getting delayed or even cancelled. You really need to have visibility into the shipment— where it is currently and where it needs to be. You also have to have the ability to instantly reroute a package in these situations to make up for lost time.
Yet, despite the growing complexity of transporting time-critical shipments, many in the industry continue to rely on manual processes. Some logistics providers will look up various flight options and pad times before flights to accommodate for unknown pickup times. Many still use the phone to coordinate drivers and carriers.
When there is a lack of visibility, mistakes can happen too, such as shipments not being transferred to the next party, packages left on the runway or delivered to the wrong address, and more. Take, for instance, an organ for transplant: a small delay can cause a missed flight which can result in a larger delay. As a result, even a 30-minute delay that does not get communicated to the transplant team could impact a patient by hours. Having the patient under anesthesia for even an extra 30 minutes increases the risk for infection and a host of other issues. An extra hour, and the organ may no longer be a viable option for the patient.
5. Has the coronavirus pandemic accelerated the need for time-critical logistics?
The pandemic put a lot of strain and uncertainty on the logistics industry. Airline schedules had never been less predictable. However, time-critical shipments in healthcare, aerospace, and manufacturing still needed to get to where they needed to go.
And even with all of the issues with flight delays and cancellations, our average delay at Airspace only increased by seven minutes in the peak of the pandemic. At the end of the day, Airspace’s technology was created to overcome industry challenges like these, and the pandemic highlighted that using algorithms to figure out the optimal path for time-critical shipments is truly the only way to determine the best and most reliable route in rapidly changing conditions.
6. How are artificial intelligence and machine learning transforming time-critical shipping?
Consider the example of a time-critical shipment traveling from Boston to San Francisco with over 180 million transport options. That package could fly out of SFO, OAK, or LAX; take a direct or indirect flight; be driven part way. While there are millions of options, only one path is optimal for biological tissue, as opposed to a manufacturing part or cryogenic materials. Now consider a package travelling from San Francisco to Hamburg, Germany; it has a quadrillion different paths it could take.
For context, it takes a computer over four hours just to count to a trillion (indicated by a one with 12 zeros after it), and a quadrillion (or a one with 15 zeros after it) is the next named number after that. Even traditional computer programs with conditional logic or graph transversal cannot solve the problem.
These computational challenges cannot be solved by humans or with other computing methods; they can only be solved by AI. Prior to AI and machine learning, about one in three time-critical shipments arrived late—that’s really unacceptable—
which is why I often say if you are not using AI, you are wasting time and money.We use AI to solve these problems because they quite literally cannot be solved without them.
When it comes to time-critical shipments, AI can be used to assess all of the possible ways to ship a specific commodity to a specific destination to determine the best route. AI looks for all of the fastest paths and uses a neural net to score them. It considers (based on all possible ways to ship an organ, and past experience doing so) the confidence level of particular routes. It can even do this for situations that it has not seen before. And it can do it in less than a second and get even smarter over time by observing and analyzing vast amounts of data.
With this information, our people can then make better decisions and have more time to focus on another important aspect—providing excellent customer service.
7. What are some common misconceptions that you wish people knew about this industry?
Many delivery estimates people hear are inaccurate. Some time-critical shipment providers say they offer 99% or higher on-time delivery on time-critical shipments. But it’s important to note what they consider to be on-time.Some may not start the clock on delivery times until they have located a driver hours later. Others may also give themselves a padded grace period.
At Airspace, we hold ourselves to a higher standard on quoted delivery times (QDTs) for time-critical shipments. The delivery time we quote starts the instant the customer places an order. It includes the time to find the driver and route the package. We don’t pad delivery times. The times we quote are the times we hold ourselves to.
Another is real-time data and real-time tracking, which are commonly used in the industry, but often do not accurately depict what is happening. Some providers still rely on manual telephone calls versus GPS to indicate status. They may also look at where the package is at a given time, but miss an important factor: is it where it needs to be at that particular time? Many use lagging, versus leading, indicators. Knowing where the package is in relation to where it should be is essential to anticipate potential delays and determine if the package needs to be rerouted to make up for lost time.
Finally, many logistics providers say they monitor humidity, temperature, and other factors surrounding the time-critical shipment. But often they are using data loggers that only show if the shipment is viable after the fact, versus in real time. This does not help the customer if the temperature rose to say 90 degrees on heat-sensitive goods at some point in travel.
8. How are you using this technology to help to advance traditional ways of doing things in the time-critical shipping industry?
A lot of companies in the logistics space say they are doing AI. But many are just using it as inspiration. We have dozens of models in the cloud that we are actively using to drive applications.
At the same time, we are not trying to automate everything. We are really passionate about having machines do what machines are good at and people do what people are good at. We want an amazing customer experience but we also don’t want people doing anything computers can do, and with AI, computers can do more and more.
The process starts when you place your order. Our patented Logistical Management System considers the items being shipped, and within seconds, automatically finds all of the flight and driver options to identify the optimal route for shipment. Traditionally this process of finding the route and driver was handled by humans and would take nearly an hour. We also generate a time and space expectation of where your package should be at any given time.
Using geofencing technology and GPS tracking to delineate specific points, we monitor these time/space expectations throughout the package’s journey. Based on this information, our system can automatically trigger system alerts and alarms for potential delays and report those to the customer in real time.
We can also monitor temperature, humidity, motion, shock, and light exposure, in addition to exact location. The entire journey of the shipment is tracked and visible in real time with shareable reports available on demand from any device. From a delayed flight, to heavy traffic, if the package is delayed, we automatically reroute it to help make up time and keep the customer updated.
9. What do you see for the future for AI and shipping?
Right now, we are solving that problem for a specific industry – time-critical shipping, but the future is applying AI in every aspect of shipping and in every industry. Once you have the ability to determine how to best optimize shipping with AI, you gain a much better expectation on when your packages will arrive.
10. How else is Airspace different as a company?
We are growing and expanding internationally. With this growth, we are focused on our people, with the goal of having the best team in the industry.
We are also focused on sustainability. We recently announced we are working with Cool Effect to help offset carbon emissions with a project focused on helping protect the Peruvian Amazon from deforestation. We achieved carbon neutrality for all of our drives in 2020.
We also draw on an amazing driver network of independent contractors we call Commanders. They are very dedicated and take their jobs seriously, knowing how important their role is in getting time-critical packages where they need to be on time.
With over 850,000 packages shipped to date, many of them organs for transplant, and other critical healthcare shipments, it’s motivating and rewarding to know we are making a difference in peoples’ lives.
For more such updates and perspectives around Digital Innovation, IoT, Data Infrastructure, AI & Cybersecurity, go to AI-Techpark.com.
Ryan Rusnak
Co-founder and CTO, Airspace
Ryan Rusnak is an experienced Chief Technology Officer with a demonstrated history of working in the logistics and supply chain industry. He has strong information technology professional skilled in Management, HTML, Technical Writing, Interviewing, and Recruiting.
