Ryanair's chief marketing officer Dara Brady said the airline's early AI efforts around upselling and personalization have not been "overly inspiring," and that the company jumped into building an AI agent for upselling "probably too soon." Speaking at Phocuswright's Travel Marketing AI Summit in London this week, Brady offered a rare, unsentimental assessment of what AI can and cannot do for a marketing organization that fills 213 million seats a year.
"The concept is far greater than the reality," Brady said, adding that there are plenty of improvements the low-cost carrier can make. The airline applies the same cost discipline and skepticism to AI investment as it does to every other line item.
The race to build a digital travel agent
Brady sees a crowded field forming around one idea: a personalized digital travel agent that handles recommendations, check-in, and flight changes. Booking.com, Expedia, and others are all chasing versions of the same product. "Our view is that if you don't create something that competes in that space, you're going to give ground," he said.
Ryanair is working on a project where an AI agent analyzes the customer base, builds offers for specific segments, and pushes those offers to distribution channels. That work currently requires a team of people. Brady said he could see an AI agent doing it "end-to-end, in real, hyper-personalization" within 18 to 24 months, with humans shifting from doing the work to overseeing it.
"It's the closest we'll get to what's been talked about for the last 15 years around personalization and hyper-personalization," he said. "We're great skeptics of that too-but I do think we can get to a granular level without human interaction."
Where AI has actually worked
Content creation has delivered real results. Ryanair now uses the same visual across social media channels in different languages, and Brady is unconcerned about occasional errors. "If there's a spelling mistake, so be it. I'd rather save €200,000 a year in translations and have an occasional wrong spelling than pay a translation company forever," he said.
That tolerance for imperfection is deliberate. "If you're looking for perfection, you'll just fall behind and get nothing done," Brady said. The same applies to AI in marketing more broadly, where AI for Marketing often stalls on quality concerns rather than shipping something usable.
Cost is another hard constraint. Brady pointed to token costs as a reason Ryanair won't let AI loose on every customer interaction. The airline gets over 75 million searches a week on its website. "If you just let that loose... you just rack up cost, and we're still only going to fill 213 million seats this year," he said. That trade-off pushes the airline toward efficiency and productivity gains in the early phases.
Decision-making: "very little" so far
Asked what AI has done to improve decision-making in the marketing department, Brady's answer was blunt: "very little." The team has moved from using AI to write snippets and summarize reports toward building insight agents that speed up data access and investigation. "That's leading us to make better and quicker decisions," he said. "But we're generally moving from people using it at a low, personal level-'how does this help me?'-into creating agents that will do tasks."
One example: an AI agent that reviews pricing history and automates quotes for group travel. Brady also said traffic from large language models remains low, and AI hasn't yet surfaced any great insight on its own.
He expects AI to become a more qualified version of search rather than a full merchant. "I don't think any of the AI vendors-Gemini, ChatGPT or otherwise-want to go all the way to becoming the merchant, taking payments, controlling the lower end of the funnel," he said. For marketing leaders, that distinction shapes where to place bets. The AI Learning Path for CMOs covers similar decisions around where AI agents end and human oversight begins.
Why this matters for marketers
Brady's assessment cuts against the usual vendor narrative. The near-term wins at Ryanair are unglamorous: translation cost savings, faster data analysis, automated quoting. The long-term bet is an agent that handles segmentation and offer distribution without human intervention. Neither requires a perfect system. The discipline is in knowing what to ship now, what to skip, and what each AI interaction actually costs.
Your membership also unlocks:
