The real AI project at your airport isn't an AI project

Agentic AI needs one coherent operational picture to act on. Most airports don't have one, and buying the AI first won't fix that.

Here's the claim: at most airports, the real AI project isn't an AI project. It's finally unifying a data layer that's been fragmented by design for twenty years. The airports doing that unglamorous work first are the ones best positioned to run agentic AI on anything that actually matters.

The reason is architectural, not something you age into with a maturity curve. A dashboard tolerates fragmentation. A person checking three different screens can mentally reconcile a stale airport operational database (AODB), a separately-updated collaborative decision making (A-CDM) feed and an operations plan last synced twenty minutes ago. They know, from experience, which number to trust and which to double-check. 

An agent doesn't have that instinct. If the AODB shows a gate as free while A-CDM's fresher feed shows the previous aircraft still on the ground, it won't hesitate the way a controller who's seen that gate run late a hundred times would. It will act, confidently and immediately, on whichever version of reality it queried first. And the failure won't look like a failure. It'll look like a normal, well-reasoned decision, right up until a ground crew is dispatched to a gate that's already occupied.

That's the sharper problem underneath the data-freshness one: these three systems aren't just out of sync, they're owned by different organizations with different incentives to keep them current. 

The AODB is typically the airport's own IT estate. A-CDM is a joint construct that airlines, ground handlers and air traffic control all feed and none of them fully owns. The operations plan sits above both, assembled from whatever each side was willing to share on whatever cadence they agreed to years earlier. 

A human bridges that gap with judgment, catching the seams because they've learned where the seams usually are. An agent has no equivalent instinct, and its errors don't stay contained the way one controller's mistake does. Ask it to act across ground handling, gate reassignment and passenger flow in a single motion, and a bad read at the base layer produces three bad calls simultaneously, each built on top of the last, before anyone downstream has a chance to notice something's off.

The industry’s leaders have already made this call, and not only in Europe. Heathrow announced a multi-year deal in March to replace its operational database and run A-CDM and its operations plan on the same platform. Melbourne Airport announced in February that its new operations platform is live, putting flights, resources and ground handlers on one shared picture, and the announcement led with that shared picture rather than with AI. GMR launched a single platform across Hyderabad’s airside, landside and terminal operations in December 2024. The early movers have decided. The open question is everyone else.

Here's the uncomfortable part, aimed at whoever's actually in the room when next year's budget gets decided. This is the least fundable-sounding project in the building. There's no demo at the end of year one. It's data migration and systems consolidation: the unglamorous work that loses every internal pitch meeting to a vendor's slick agentic demo. The one that shows an agent rebooking a delayed connection in real time, projected on a screen in front of your board. 

That demo will keep winning the budget fight, because it's built to, and because the honest counterargument, that you need eighteen months and several million dollars to reconcile three databases nobody outside operations has ever heard of, sounds like an excuse dressed up as a project plan.

But it is  the actual prerequisite, and the airports that skip it aren't choosing a faster path to the same destination; they're choosing a destination that looks identical in the demo and falls apart the first time the agent is handed something that actually matters. 

The airports willing to fund the unglamorous work are running ahead on the one investment that was never optional to begin with. They just recognized its importance before their board made them prove it the hard way.

It’s worth looking at the airports everyone points to. Singapore Changi, Incheon and Hong Kong are three of Skytrax’s 2026 top four, and they’ve earned it: biometric journeys, service robots, 5G-enabled autonomous vehicles, immersive terminals built to be photographed. 

But a ranking measures what passengers experience. It doesn’t tell you whether the data underneath is coherent enough to hand an agent real operational authority, and nothing in how these three publicly describe their digital programs says they’ve made that move. Munich, ninth on the same list, has reportedly adopted the same kind of platform, so a top ranking and a coherent data layer can coexist. They’re earned by different kinds of investment, and only one of them comes with an award.

The question worth asking in your own next planning cycle is therefore whether anyone in the room can currently give a straight, confident answer to what your airport's own operational picture actually is, right now, without three people checking three different screens first.