Malaysia Airlines has deployed an AI customer service agent named Mavis, built on the agentic customer experience platform from Ada, a Toronto-based automation company, the carrier and Ada announced on February 24, 2026. Rather than a scripted chatbot that routes nearly every question to a human, Mavis is designed to resolve requests on its own by reaching into the airline’s live operational systems, checking actual bookings, confirming real gate numbers, and processing check-ins — a distinction the companies say sets it apart from the FAQ-style bots that have long frustrated airline passengers.
A Bot That Touches Real Bookings
According to Ada’s announcement, Mavis operates 24 hours a day across Malaysia Airlines’ website, mobile app and email, handling what the companies describe as high-value, time-sensitive inquiries: flight status and schedule changes, booking and itinerary lookups, check-in access, boarding gate information, lowest-fare discovery and seat upgrades. The system currently supports English and Malay, with additional languages planned to serve the carrier’s international customer base. Because Mavis is tied directly into Malaysia Airlines’ operational data rather than a static knowledge base, it can, for instance, tell a passenger their actual gate assignment rather than generic information about how gates are typically posted.
Part of a Broader Industry Shift
Malaysia Airlines’ move lands amid a wider push by airlines to automate customer service with agentic AI. Delta Air Lines has begun rolling out its own generative AI-powered virtual assistant, and India’s SpiceJet said in early October 2026 that AI “has done extremely well” for its customer operations, with its chief customer officer noting that many customers cannot tell whether they’re talking to a bot or a person. Industry data cited by social-media analytics firm Emplifi suggests AI now resolves close to a third of airline customer conversations on platforms like Facebook without any human involvement, and airlines globally are automating more than a quarter of direct-message customer interactions.
Why Airlines Are Betting Big on Agentic AI
The appeal for carriers is straightforward: airline customer service is notoriously volume-heavy and spiky, with disruptions like weather delays or mechanical issues generating floods of nearly identical questions about rebooking and refunds within minutes. A system that can resolve routine inquiries without a phone queue reduces wait times and call-center costs simultaneously. Ada’s platform is pitched specifically on the idea of full resolution rather than simple deflection — meaning the goal isn’t just to answer the easy questions and shunt everything else to a human, but to actually complete transactions like seat changes end to end.
Why Malaysia Airlines, Why Now
The airline’s decision to lean into agentic AI comes as Malaysia Airlines Group, which has spent years working through a turnaround plan after financial troubles in the 2010s, looks for ways to raise service quality without proportionally growing its call-center headcount. A carrier serving long-haul and regional routes across Asia, the Middle East and beyond faces a customer base with widely varying language needs and expectations, which is part of why the company and Ada have emphasized multilingual expansion as a near-term priority rather than a future aspiration. Positioning Mavis as a system that resolves issues rather than merely deflects them is also a direct response to years of customer complaints across the industry about chatbots that loop travelers through menus without ever solving their problem.
The Case for Caution
Passenger advocates and some aviation consultants remain skeptical that agentic systems are ready to handle the messiest parts of air travel — misconnections during irregular operations, disputes over compensation, or passengers with accessibility needs who require nuanced handling. An AI agent empowered to directly modify bookings or process check-ins also raises the stakes of a wrong answer: a misread itinerary or an erroneous gate change pushed to a confused traveler carries real consequences in a way a wrong answer from a static chatbot does not. Malaysia Airlines and Ada have not published independent accuracy or escalation-rate figures for Mavis, and the companies’ own materials acknowledge the system is designed to hand off to live agents “when human support is needed” — without specifying how often that threshold is triggered in practice.
What’s Next for Mavis
Malaysia Airlines says it plans to expand Mavis into voice channels and add “agent-assist” capabilities that would support human staff handling harder cases, along with a dedicated itinerary-builder function. The airline is also expected to broaden Mavis’s language support beyond English and Malay as it scales the system across its international routes. For an airline whose customer base spans dozens of countries and languages, the real test of Mavis won’t be how it handles a routine fare lookup — it will be how gracefully it steps aside, and how quickly, when a passenger’s situation is too complicated for an algorithm to untangle alone.
Photo: ClickerHappy / PIXABAY via Pixabay
