Pod-Ed: Why Airline AI Needs Better Data Before Smarter Technology

NEW YORK — For Tim McMullen, the promise of artificial intelligence in aviation depends on the data supporting it. New applications must work with the older systems that airlines use to handle bookings, schedule crews, and manage their operations.

In Season 10, Episode 8 of The Airways Podcast, hosts Rohan Anand and Vinay Bhaskara welcome McMullen, vice president of aviation and travel technology at DataArt, to examine the realities of airline technology modernization. Drawing on his airline and software-industry experience, McMullen explores why airlines struggle to integrate legacy platforms, how AI is changing software development, and why passenger experience remains a critical measure of technological progress.

Airline AI Is Only as Good as Its Data

McMullen describes airlines deploying AI-powered applications while continuing to operate foundational systems developed decades ago.

He and Bhaskara describe technology stacks spanning different eras: passenger service systems, crew scheduling platforms, loyalty databases, and mobile applications. McMullen traces the origins of passenger service systems to the 1960s; Bhaskara emphasizes the work required to connect different database structures.

For McMullen, integrating these systems represents one of the industry's most important challenges.

"The secret sauce of AI is really that data platform."

He argues that companies frequently underestimate the importance of preparing their data infrastructure before launching AI initiatives. Without dependable information, consistent business rules, and interoperability between systems, even promising applications risk stalling before reaching production.

Bhaskara points out that airlines are particularly vulnerable because their technology stacks have evolved unevenly. A modern customer-facing application may depend on operational information stored in systems built using entirely different database structures.

Modernization Without Replacing Everything

Airlines cannot simply switch off their existing technology and start again.

McMullen highlights passenger service systems from Sabre and Amadeus as examples of established infrastructure that he considers reliable. He says airlines must also manage the risk of older applications unexpectedly going out of service.

His alternative is selective modernization: airlines can develop new applications that extract and interpret information already contained within established systems. He says building a new passenger service platform would fall outside DataArt's preferred work when existing products already perform that function.

McMullen points to AI assistance at United Airlines (UA), Delta Air Lines (DL), and American Airlines (AA) as examples of newer tools alongside older passenger service and loyalty systems. His point is the orchestration required behind the scenes, rather than a technical assessment of any one airline's application.

For passengers, the result might appear as a simple notification, boarding update, or personalized recommendation. Behind the scenes, delivering that experience requires multiple systems to exchange accurate information.

The Difference Between Technology and Personalization

The hosts explore how airline technology priorities differ depending on an airline's business model.

Bhaskara draws a distinction between low-cost carriers seeking a consistent, efficient, predictable operation and premium carriers pursuing more individualized experiences. McMullen says they may use similar tools, deployed with different priorities.

McMullen recalls reaching a lifetime frequent-flyer milestone with a major U.S. airline without receiving acknowledgment during the journey.

He says the occasion passed without recognition. He would have valued an acknowledgment from a Gate Agent or Flight Attendant, even without an upgrade. The anecdote does not establish why the airline missed the opportunity; for McMullen, it illustrates a use for information already held in a loyalty program.

Later, he proposes extracting milestone information from a loyalty application and passing it to frontline employees. That future-flight example is hypothetical, while the earlier missed recognition is his own recollection.

The discussion then turns to ZIPAIR (ZG), a wholly owned subsidiary of Japan Airlines (JL). McMullen describes its “new basic” positioning, and Anand calls it a value carrier.

McMullen describes DataArt's work developing a payment orchestration platform for ZIPAIR. DataArt's case study describes the platform and its support for multiple payment methods.

As McMullen explains it, the objective is to reduce friction during booking by giving customers payment options appropriate to their location.

He uses ZIPAIR's flying between Japan, other Asian markets, and North America to explain why payment preferences matter. In his account, payment flexibility is part of the passenger experience alongside the onboard product.

Southwest Airlines and the Cost of Technical Debt

Anand brings his experience at Southwest Airlines (WN) into the discussion, examining how technology decisions made decades earlier can constrain an airline's commercial strategy.

Anand describes Southwest's historically simple operating model and argues that its technology reflected those priorities. In his account, a focus on low costs and efficiency eventually had to accommodate a more complex business.

He recalls technical limitations becoming more consequential as Southwest expanded internationally and explored partnerships, ancillary revenue opportunities, and more sophisticated distribution.

The conversation revisits Southwest's acquisition of AirTran Airways and the different capabilities of their systems.

Anand recalls how the integration of different reservation and distribution platforms demonstrated the commercial potential of technologies that had previously been difficult for Southwest to support.

Anand's argument extends beyond Southwest: technology architecture can enable or restrict an airline's commercial ambitions. Decisions that appear sensible for one business model may become liabilities when that model evolves.

McMullen argues that successful modernization therefore requires coordination between technology leadership and the wider business.

Data Sharing, Partnerships, and Passenger Trust

McMullen describes data sharing as part of the passenger experience across codeshares, interline agreements, joint ventures, and other partnerships. He uses seat assignments across oneworld airlines to explain the difficulty: a passenger sees one booking, but selecting a seat can require separate reservation platforms and seating systems to exchange information.

Sharing passenger records also raises questions about ownership and security. McMullen extends the discussion to airport biometrics and asks which organization owns the data, who secures it, and whether a partner should retain it once its intended use is complete.

For him, cooperation between partners needs to be supported by governance and clear rules. The aim is to give each partner the information it needs to serve the passenger, with limits on how that information is used and retained.

From AI Experiments to Working Applications

McMullen says DataArt works with airlines to identify where AI could be useful and where a different solution would suffice. He describes a workshop involving crew scheduling, dispatch, maintenance, finance, marketing, and airport customer service. The team organized the resulting ideas into immediate improvements, follow-on projects, and larger strategic investments. Some depended on additional data infrastructure; others did not require AI.

In a separate example, McMullen says DataArt automated a longstanding manual process for an unnamed client in about nine weeks, including a handoff involving other technology vendors. He says the work saved 83% of the time spent on that process. The client is not identified, and the figure is his account of the project rather than an independently verified result.

AI is also changing how the software itself is built. McMullen describes agents generating code, finding defects, and proposing or implementing fixes. He says humans still have a substantial role in understanding business requirements and does not regard agents as ready for every case.

Aviation Remains a People Business

Despite the emphasis on automation, the episode closes with a discussion about the human consequences of technological change.

Anand raises concerns about employment, institutional knowledge, and the importance of responsible AI adoption. McMullen acknowledges that some businesses may need fewer employees but argues that experienced airline employees remain valuable because of their institutional knowledge. He hopes customers retain those employees and says DataArt does not determine their staffing decisions.

He also stresses the importance of working directly with customers rather than assuming technology can replace every human interaction.

Airline operations are complex, and understanding how employees and passengers actually use systems often requires firsthand observation and collaboration.

For McMullen, the objective is not simply to automate more tasks. It is to help airlines identify where technology can improve their operations while preserving the experience and expertise needed to make those improvements meaningful.

What the Discussion Suggests

The discussion makes a case for treating integration, governance, and execution as central to airline AI adoption. McMullen's data-foundation argument connects Bhaskara's account of mixed-era systems with Anand's concerns about commercial strategy and responsible adoption.

The participants describe reservation systems, loyalty platforms, operational databases, and customer-facing applications designed around different business requirements.

Those systems contain an enormous amount of potentially useful information, but their value depends on whether airlines can connect that information reliably and make it available where decisions are actually made.

The commercial implications are substantial. An airline that cannot connect its loyalty data to frontline operations may miss opportunities for personalization. A carrier whose reservation platform cannot support new distribution models may struggle to execute its broader network strategy. And an AI application built on incomplete or poorly governed data may introduce complexity without delivering measurable improvements.

Their examples suggest that dependable data infrastructure, disciplined modernization, and leadership connecting technology to the business will shape what airlines achieve with AI. Their concern for institutional knowledge and human contact also sets a test for those investments: whether they improve the work and the passenger experience people actually encounter.

An earlier Pod-Ed explored how technology and distribution shape an airline’s strategy.

The Airways Podcast is hosted by aviation analysts Rohan Anand and Vinay Bhaskara and produced by Airways Digital Editor Helwing Villamizar. Watch Season 10, Episode 8 with Tim McMullen of DataArt discussing airline technology modernization and the future of AI in aviation below.