Generative AI ‘Super Analyst’ May Change How Airlines Set Prices

Airlines are broadening their use of artificial intelligence to unlock efficiencies, profits and a more tailored customer experience. One application gaining traction lies in pricing, with active trials poised to evolve how carriers sell.

Eight airlines, six publicly, have begun tests with Fetcherr, an Israeli startup offering a pricing engine driven by generative artificial intelligence (GenAI). The company’s first U.S. customer is Delta Air Lines. While the carrier generates nearly half of the domestic industry’s profits, it sees room for improvement, so it is assessing a “full reengineering” of its pricing system, Delta President Glen Hauenstein said in November at an investor day.

• Advancements in AI provide new opportunity for airlines
• Pricing, maintenance and customer service tools are being tested

At that time, about one year into a multiyear, multistep partnership, Delta was pricing 1% of its network with Fetcherr’s GenAI technology. “The initial results show amazingly favorable unit revenues versus the beta,” Hauenstein said, describing the potential for prices tailored to flight, time and individuals. “We’re all in on this,” he said.

Fetcherr’s five other public customers are Azul, Royal Air Maroc, Viva Aerobus, Virgin Atlantic and WestJet. Founded in 2019, the company chose initially to specialize in the airline industry, with its three co-founders coming from algorithmic trading, e-commerce and advertising. Fetcherr sought out “the right customers willing to, let’s say, break the glass ceiling of legacy,” CEO and co-founder Roy Cohen explains.

“Me and my partners, we are not aviation people. . . . Our connection to aviation and travel is as any normal customer who is buying a ticket,” Cohen quips. “We came up with the notion of our large market model without considering aviation because technology is agnostic. It can work in any industry.”

For Fetcherr’s launch, “we were looking for industries that are extremely legacy so we can showcase how our GenAI [large market model] can optimize the way companies or large organizations price,” Cohen says. “We saw a big opportunity there [with the airlines] to bring a technology in that can basically future-proof the way these organizations work. At the end of the day, it’s beneficial for the end customers because the prices are more flat throughout the booking calendar.”

At the heart of Fetcherr’s offering is a large market model, an AI engine able to forecast demand and market trends, learning as it goes. A demand-based system, it takes in airline schedules and inventory, public data from vendors including OAG, and weather and financial market data, acting as a “super analyst” to optimize the price until departure.

“The system recommends what to do, and the [airline] analysts can always overrule the AI,” Cohen explains.

Although airline moves to test AI in pricing already have come under U.S. government scrutiny, carriers have pushed back, describing the focus as being on efficiency and optimization. A December U.S. Senate subcommittee hearing on ancillary fees questioned representatives from five airlines on their plans for the technology, seeking to understand whether AI use for that purpose could be exploited and harm consumers.

Delta Chief External Affairs Officer Peter Carter countered at the hearing that the AI push is “about the right offer at the right time, as opposed to a thousand different offers to a thousand different people.”

Fetcherr stresses that it does not collect any customer’s private data.. “We don’t price customers; we price fares,” Cohen asserts. “We have zero private information.”

Once implemented, Fetcherr says its GenAI solution can boost revenue 6-9%. Virgin Atlantic, its first customer in the UK, already has seen an impact, even if AI output initially had to fit in with legacy systems during a transition phase.

“It isn’t about cost saving for us, but it does change how we work,” Chris Wilkinson, vice president of pricing and revenue management at Virgin Atlantic, said during a podcast hosted by Fetcherr. “It’s there to help you perform better, to do more.”

Virgin set out to measure the technology’s effectiveness by comparing new and old ways of pricing.

“There’s no human putting a step between AI making a recommendation and it going into the market,” Wilkinson explained. “Instead, we put our resources after the outputs have gone in to understand why it did what it did. We gave half of a route to Fetcherr, and we kept our existing systems and processes for the other half and then measured one versus the other.”

As the airline has moved out of a proof-of-concept period and into implementation, “we think we’ve seen the benefit grow over time,” Wilkinson said. “We think it keeps on learning.”

BEYOND PRICING

AI is not new to airlines. Traditional AI, which can analyze and predict, has been used for functions including revenue management, marketing and network planning. But the emergence of GenAI—capable of creating new content based on what it learns—is broadening how the ever-evolving technology is deployed. And while airlines investigate its potential to shake up pricing, other operational areas are also benefiting.

One example is EasyJet, which opened a new operations Integrated Control Center (ICC) in Luton, England, to manage its daily program of about 2,000 flights, embedding AI into its day-to-day practices to speed up and improve decision-making.

The ICC, which opened in May, houses 250 specialists and operates 24/7, using AI for predicting standby crew requirements, including a crew planning tool that helps recommend and select the best crew options for the needs of EasyJet’s operation and aircraft.

The new GenAI tool, Jetstream, gives the team “instant access to policies, procedures and information which will enable them to solve operational issues as they occur,” EasyJet said in a press release at the time, noting that it planned to place AI-led technology in crewmembers’ hands as well.

Air France-KLM also is using AI to improve its operations and offerings to passengers. In December, the group and Google Cloud announced a strategic collaboration. Through this greater use of data and AI, including GenAI, Air France-KLM wants to understand passenger preferences and travel behavior better to provide tailored options and services and to improve operations including through predictive aircraft maintenance.

Air France-KLM has begun transitioning its three legacy data centers toward a multicloud strategy using Google Cloud’s data and analytics tools, including Big-Query. The group calls the partnership with Google Cloud “a significant step forward” in its data strategy.

“Airlines generate massive amounts of data, much of which can be incredibly valuable in helping drive operational insights, build better customer experiences and—with the power of GenAI—create entirely new services and offerings,” said Matt Renner, president of global revenue at Google Cloud.

Even before the collaboration with Google Cloud, Air France-KLM was exploring the potential of AI to improve operations and offerings, as fleet renewal brings into play aircraft that communicate increasing amounts of data to their operators.

In early 2023, the group decided to embrace ChatGPT and see what it could offer, Julie Pozzi, head of operations research, data science and data strategy, said at a May briefing in Paris. “There has been a strong acceleration in recent years,” Pozzi noted. “We have been focusing on data strategy, and we have launched data management programs to structure the way we manage data at Air France and KLM.”

Conscious that GenAI is in its infancy and technologies that seem the latest thing today might be obsolete in a year’s time, Air France-KLM is not developing its own solutions but is basing its GenAI experiments on existing market solutions.

While some 40 GenAI projects were underway at the time of the briefing, a select few were at the proof-of-concept stage, being tested or in use, Pozzi said, including Air France’s own internal ChatGPT, Talia, which allows employees to familiarize themselves with the tool in a closed circuit. Another tool, Pamelia, provides Air France airport staff with quick answers to questions from customers via their tablets. One big advantage is the voice function, which can answer the customer in 85 languages.

Another GenAI tool, Charlie, aims to improve maintenance processes by quickly finding part numbers within reams of documentation and thereby speeding up the replacement process—good for the airline’s punctuality stats and for passengers waiting on the tarmac as well as for technicians’ stress levels, Air France said.

Fox, meanwhile, is a tool capable of automatically analyzing customer feedback—including picking up on humor—and ideal for helping the airline zoom in on passenger opinions of a given aspect of their journey, such as onboard catering.

“Generative AI is not really disruptive; it is more a continuation of the data transformation of the airline,” Pozzi said, noting that it is suitable only for certain use cases—“the cherry on the cake.”

AI also has a role to play in cybersecurity; carriers are using the technology to help safeguard their operations. Latvia’s AirBaltic says it deploys multiple integrated AI tools for continuous monitoring of its network and applications, allowing it to identify and respond to potential threats in real time.

The carrier also is using AI-based technology to maximize ancillary revenue, including PROS’ Dynamic Ancillary Pricing and Merchandising, which allows for personalized offers based on detailed customer segmentation. The system, introduced in late 2023, automates and optimizes the pricing of seat assignments, allowing AirBaltic to adjust prices dynamically based on demand and customer preferences. Customer interactions help determine the optimal price points for various segments and flights, enabling the AI algorithms to learn and improve predictions over time.

AirBaltic describes the results as promising and is considering extending AI-driven strategies to other ancillary products such as baggage.