How Air France is leveraging generative AI for next-gen customer service

Source: Supplied

Air France's commitment to data-driven solutions began in 1958 with the establishment of a dedicated operational research department. This department focused on innovation and adapting the company to the ever-changing aviation industry. Recognizing the potential of customer data early on, Air France leveraged data analytics to expand its revenue management strategies in the 1990s, utilising historical information and trend analysis.

This focus on big data continued into the early 2000s with the development of Prognos, a predictive maintenance solution utilizing large data sets for scenario planning in the maintenance business. Prognos marked Air France's entry into the era of predictive and prescriptive artificial intelligence, a technology now used by over 80 airlines worldwide.

Since then, artificial intelligence has become a cornerstone of Air France's research and innovation programmes. Today, AI is integrated throughout the customer journey, with applications ranging from chatbots to tools that predict the number of bags and meals on board, calculate water needs, and optimise flight paths for fuel efficiency (eco-piloting).

These diverse uses of AI all share a central goal: optimising operations, anticipating customer needs, and providing staff with easy access to job-relevant information. Ultimately, this data-driven approach with AI at its core aims to deliver an enhanced customer experience.

2023: A generative leap

Since 2023, Air France has embraced a new revolution in artificial intelligence: generative AI. This technology differs from traditional AI by its ability to independently generate rich content. It relies heavily on machine learning, essentially feeding on and learning from all user interactions.

Generative AI, popularised by tools like ChatGPT, represents a significant technological advancement and opens exciting new avenues for data exploration.

Air France has launched over 80 projects utilising generative AI across various business sectors. These projects are at different stages, with some identifying the most suitable solution (data management, predictive AI, etc.) and others undergoing proof-of-concept testing.

Four examples of Air France's AI advancements:

• Talia: Air France's internal ChatGPT equivalent, allows employees to learn about the technology in a secure environment without sharing information with third parties. Employees use TALIA daily for tasks like writing emails, searching PDFs, organising events, and creating to-do lists.

• Pamelia: This solution empowers airport agents to answer customer questions directly on their iPads. Pamelia searches Air France's reference manuals and procedures to find answers to queries on baggage allowances, live animal transport, and formalities. It then generates written responses, translatable into 85 languages, for immediate customer communication. Currently in testing, Pamelia is slated for deployment at Paris-Charles de Gaulle in 2025.

• Charlie: This tool aids maintenance teams by searching aircraft parts in both airline and manufacturer documentation. Charlie saves valuable repair and replacement time, contributing to improved flight punctuality.

• Fox: Designed to analyse customer feedback, Fox utilises generative AI to automatically analyze diverse and complex texts, even those incorporating humour or irony. It helps Air France understand customer expectations and concerns, identify emerging trends (weak signals), and share this information across the company.

Air France's approach to artificial intelligence prioritises responsible development. They focus on:

Data protection and ethical use: Customer and company data security is paramount, along with adhering to regulations and ethical principles. Solutions are kept within closed circuits, and an AI committee ensures these principles are upheld.

Supporting internal initiatives: Their 150-person operational research team assists various departments in exploring AI possibilities. The team can recommend and guide on project complexity, support proof-of-concept development, and participate in implementing chosen projects.

Leveraging existing solutions: With AI, especially generative AI, still being a relatively new field with evolving technologies, Air France prioritises using established market solutions rather than developing models in-house. This approach helps control investment costs and maintain agility by avoiding quickly outdated technologies.