With the meteoric rise of generative AI chatbots like ChatGPT, it’s very possible that 2023 will be looked back on as the year that artificial intelligence came of age.
But while AI may have captured the attention of the public and industry alike, there is still some way to go before it becomes fully embedded in business aviation, says Steve Varsano, founder and CEO of London-based aircraft broker The Jet Business.
“I do believe the letters AI are thrown around a lot today and have huge potential to control a lot of industries. However, it may take a little longer for existing computer and programming technology companies in the aviation space to really focus on the benefits that could come from AI,” says Varsano, who is currently researching the opportunities offered by the technology.
In fact, AI has already found its way into private aviation, with a number of software developers and private jet operators offering AI-powered solutions to improve operations and enhance customer experience.
The German on-demand regional air mobility startup Flyv, for example, leverages proprietary AI algorithms as part of its efforts to develop a new generation of custom-made flight networks.
Elsewhere, Stack, which offers business operating systems for charter brokers and operators, also uses AI in its software tool.
AI pioneers
One of the earliest adopters of AI in the private aviation realm was MySky, a Switzerland-based company that developed an AI-based online audit platform for private jet operators back in 2016.
Chris Marich, MySky CSO and co-founder, says the idea for the company came after observing the technology deficit in the management systems used by private aviation.
Marich and his co-founder Kiril Kim found themselves asking the question “how come these aircraft, which are so technologically advanced, are managed in such an old-school way?”
Marich says, “Looking at the industry from the outside, everyone thinks that everything is optimized and great, but actually we realized that a lot of people were using Excel spreadsheets and Post-It notes to manage US$100-million assets, which didn’t make a lot of sense to us. So that’s when we started thinking about the use case for our product.”
While the MySky product has “evolved over the years” to a platform focused on spend and revenue management optimization, the starting point for the company was creating an automated solution for invoice verification, says Marich.
“We were asking ourselves: why do we still invest so much manual and human capital into some very time-consuming and repetitive tasks? How can we use technology to automate the data ingestion, verification, breakdown and customization? That’s really where we started. Our industry is unscheduled aviation so you receive a lot of invoices from a lot of different suppliers. There is no standard, there is no format, there is no comparison – and there is no benchmark.
“So the invoice can arrive in a lot of different formats and layouts and every airport has different rules – from one airport there might be three line items on the invoice, from another airport there might be 30. There is zero consolidation on that.”
Marich and his team developed an AI algorithm capable of sorting through and verifying all these myriad types of invoices. The AI does this in two stages. The first stage involves analyzing line-by-line all the invoices and allocating to each expense an aircraft, an operator and a trip number.
Once this pre-accounting work is done, the AI moves on to the second stage of automatic verification.
Marich explains that “the machine cross-checks operational data with financial data and can easily detect mistakes” in the invoicing, for example, if an account has been double-charged.
According to Marich, AI is a cornerstone of MySky’s technology, which he says was “the first AI-powered spend management platform in the industry.” He says, “The AI allows us to do tasks that are very repetitive, that a human being is likely to make mistakes with because they get bored.”
But it is not simply that the AI alleviates human boredom. The speed at which the AI is capable of sorting through large quantities of data vastly outpaces any human attempting the same task.
This ability to process large and complex datasets is one of the biggest advantages offered by AI, says Vinay Roy, senior vice president of product at Vista Global Holding, a Dubai-based private aviation group that owns the on-demand private jet company XO.
XO uses AI to underpin its mobile app, which provides users with an immediate view of aircraft availability and pricing across a digital aviation marketplace of 2,100 private jets, which includes Vista member’s own fleet of more than 360 aircraft.
“Private aviation operation is becoming increasingly complex because of the need to operate in an unscheduled charter flight environment while also delivering highly complex trips from diverse private terminals,” says Roy.
Updating real-time data on Vista’s own fleet, which operates in more than 190 countries globally, requires grappling with millions of potential routes in multiple times zones near-instantaneously.
“No human system can scale to solve this problem,” says Roy.
According to Roy, by deploying AI, Vista’s in-house management system is able “to track the entire Vista fleet of aircraft, providing full visibility into aircraft availability, maintenance, and schedules.”
Roy adds, “Underneath the software is a layer of augmented/ artificial intelligence, which is leveraging hundreds of data points and machine-learning algorithms.” All this AI is taking place behind-the-scenes, and is invisible to the end user.
This is one of the primary differences between the AI developed by XO and MySky and the generative AI tools that have drawn so much public attention over the last year. Chatbots like OpenAI’s ChatGPT are designed to interact directly with the end user, making the technology much more visible.
Whereas in fact, most of the AI applications used across industries today are embedded into systems and solutions where their function is largely invisible to the end user. The high visibility of ChatGPT is one explanation for the software’s success. The other is the software’s impressive creative capabilities.
Using data gleaned from online and other sources, ChatGPT, and chatbots like it, can generate new data – hence why they are called generative AI. They can compose poems, write essays on obscure subjects, even replicate the voices and musical styles of famous artists.
Impressive as all this can seem, finding a practical application for generative AI in the private aviation space remains a work in progress, says Marich.
“Right now, everyone is trying to put the technology everywhere to look cool,” says Marich.
“We decided to take our time. Instead of implementing it just to create a wow effect that has limited benefits, we’re investigating where this brings concrete value to our customers.”
Generating data
MySky’s AI-powered offering now also includes pre-flight estimates, on top of the post-flight management for which the software was developed. The success of their AI offerings is centered on the so-called virtuous circle approach in which the primary resource is always data.
“The more data you have the better it works, and the better it works the more people you attract,” says Marich. “And the more people you attract, the more data you generate, and so on.”
Vista are also looking at where else they can integrate new AI technologies like generative into their portfolio. But finding other areas where AI can play a role in aviation “is never going to be a binary zero or one,” says Roy. “I think the right approach is to look at an augmented intelligence model in which there is a close handshake between machines and humans.”
In such a model, he says, the AI can be left to do tasks where it excels – for example, repetitive tasks – while humans can retain control of tasks that require decision-making intelligence.
Moreover, in an industry like private aviation where customer service plays such an integral role, the human factor is not something that can be easily dismissed.
“Human interaction, at least for the time being, is hard to replace because people still get frustrated when they cannot speak with a person to solve their different issues,” says Varsano. “Generally, people have become much more impatient and just feel more attended to when speaking to a human.”
Marich agrees. “We are in an industry where human contact is key,” he says. “This will never be able to be replaced by a software engine.”
