Inside Kenya Airways’ AI Playbook

For an airline operating on some of the thinnest margins in global aviation, the case for AI at Kenya Airways was never going to be made on novelty. It was made on arithmetic.

“The margins in Africa are less than 2.5 dollars net profit per passenger,” Fred Kitunga told CIO Africa on the sidelines of the AWS Summit in Johannesburg, when asked why the airline is moving from AI experimentation into full deployment. Elsewhere in the world, he said, airline margins run “upwards of 15 dollars net profit per passenger, and sometimes even 20 dollars net profit per passenger.” That gap, more than any enthusiasm for the technology itself, is what is driving the pace. “We must use all available channels and capabilities to make sure that we create value,” he said. “It is about making sure that we accelerate that value.”

For Kitunga, the mandate from leadership is unambiguous: keep the airline operational and profitable, and use AI to get there faster. That means looking closely at customer lifetime value and ancillary sales opportunities generated through the airline’s many touchpoints with travellers — and organising those opportunities into what he called “a structured approach.”

Working Backwards From The Problem

Kenya Airways’ approach to deploying agentic AI is deliberate. Rather than starting with the technology and looking for places to apply it, the airline — with AWS as a partner — works in the opposite direction. “We do not begin with a preconceived answer,” Kitunga said. “We first identify the need and then apply the appropriate AI Tool for use cases.”

That discipline shows up across three areas where the airline says it is already seeing measurable value.

The first is internal: employee productivity Amazon Q has been rolled out to more than 1,000 employees to support day-to-day work. The effect, according to Kitunga, has been a compression of task time: activities that “previously took days or weeks” are now measured in hours.

Kenya Airways has also been at the forefront of training and building AI capability among its staff, recognising that successful AI adoption requires not only the right technology, but also people who are equipped and confident to use it —  partly, Kitunga said, to help employees feel the technology is working for them rather than around them, in what he described as a broader cultural shift “from very orthodox and traditional ways of doing business towards relying on capabilities that are automated.”

The second area is customer service, where Kenya Airways draws on a central system that consolidates customer data to identify individual travelers who have experienced disruptions — sometimes before those customers have reported the problem themselves. ” We are able to proactively identify these issues before a customer complains, and resolve them before they escalate” he said.

The third is maintenance, where AI is being used to accelerate a historically manual, document-heavy process. “There is a lot of manual reading involved,” Kitunga said. “Maintenance teams have to review large amounts of data and compare one set of information with another.” AI, he said, speeds up the identification and resolution of maintenance issues on our aircraft.

Governance Before Deployment

Aviation is among the most heavily regulated industries in the world, and Kitunga was direct about the trust problem AI still has to overcome, both internally and with regulators. “AI has to be explainable,” he said. “There must be explainability in how AI is used.”

His response was to formalise governance ahead of deployment rather than after it. Kenya Airways developed a dedicated data and AI policy that went to the board for approval, informed by practices already in use elsewhere in the industry. Every AI use case is now assessed against that policy before it proceeds. “We do not use AI simply for the sake of using it,” he said. “Every use case must fall within the policy.” He linked the approach directly to the airline’s broader safety culture. “We are a very safety-conscious airline. We believe that safety is central to the delivery of our business, so everything is based on that principle.”

The Data Problem Behind the AI Problem

Asked how much of Kenya Airways’ institutional knowledge — flight data, customer history, regulatory filings — was actually accessible to build AI capabilities on top of, Kitunga did not overstate the airline’s starting position. “To be honest, that is still a challenge,” he said. “It was a challenge before, and we remain conscious of it.” Data across the organisation, he said, remains “segmented, siloed, unstructured and spread across different parts of the organisation.”

The response has been to build a data lake, using Amazon Bedrock to map and consolidate scattered data sets, remove inconsistencies, and establish a unique identifier per customer. The goal, Kitunga said, is a single, secured source of truth that can then be used in production — a project he described as still under construction rather than complete.

Where AI Can’t Go Yet

Asked which process he believes could benefit most from AI but has not yet been able to deploy it, Kitunga pointed to flight simulation and flight data — “the entire journey and how we determine the efficiency of a flight, including capacity and other operational factors.” In principle, he said, it is an area well suited to automation. In practice, Kenya Airways remains bound by dependencies on original equipment manufacturers, whose manuals and established procedures maintenance engineers must still follow, along with restrictions from IATA and other aviation authorities. “The capability has also not yet been fully developed for this purpose,” he added.

Customer data presents a related but separate constraint. Kenya Airways holds large volumes of customer information, but its use is governed by compliance obligations, including the General Data Protection Regulation (GDPR). The limiting factor, Kitunga said, is consent. “If a customer has not provided consent, we cannot do anything with that customer’s data.” The airline is now working to educate customers and secure that consent progressively, describing it as a bottleneck to wider adoption rather than a closed door.

The Partnership, And The Advice

Kitunga was direct about the role AWS has played, describing the selection of a technology partner as a deliberate decision made against a crowded field of AI providers. “The partnership has helped us transform our airline,” he said. “It has shown us that activities that would previously have taken too long to implement can now happen much faster.”

Asked what advice he would offer other organisations starting down the same path, his answer returned to the same principle that has shaped Kenya Airways’ approach throughout. “Do not begin by thinking about which tools to use. Begin by identifying the problem you want to solve, then work backwards to find the right solution.”

This interview was conducted at the AWS Summit in Johannesburg.