From AI talk to AI Impact. What I’m hearing out there

Spring conference season is done and I’m finally back at my desk. The last few months have been a blur of early flights, good conversations, and the kind of energy you only get when an industry is genuinely wrestling with something. AI in airports is that something right now, and the conversations have shifted noticeably even in the last year.

Here’s what’s coming up consistently.

The Business Case for AI in Airports Is More Honest

The theoretical excitement around AI hasn’t gone away, but it’s sitting alongside a much more grounded conversation now. Airports are under real pressure: the desire to grow revenue per passenger, rising labor costs, and experience expectations that keep climbing. The response has often been to add people, more staff, more visible effort. It helps, until it doesn’t, and everyone in these rooms knows it.

What’s changed is the question being asked. Not “what could AI do?” but “what actually holds up six months after go-live, in a real airport, on a disrupted day?” That’s a harder question and a much more useful one.

Data Is Where the Friction Lives

A lot of the operational AI conversation starts (as it should) with data: where it comes from, who owns it, and what happens when priorities shift.

Passenger flow: measure what matters

On people flow, the sensor debate (camera versus LiDAR, repurposing CCTV, edge processing) is active, but it’s almost the wrong conversation to have first. The more important question is what is being measured and how useful it is.

Knowing the length of a queue that formed twenty minutes ago is interesting. Knowing the predicted wait time for the passenger joining the back of it right now is operationally useful. And when that TSA queue spills past the stanchions and into the hall, can the installed system still see it? Check-in areas add their own complexity: systems that simply measure queue length aren’t operationally useful.

The same story came up repeatedly. Some initiatives were undone when a CCTV camera was moved or pointed elsewhere for security purposes, which broke the other uses that camera was supporting. Others fell short because coverage and accuracy commitments made in a demo didn’t hold up in a full deployment, in a real terminal on a real day.

What airports do with the data

Another interesting thread is what airports are doing with that data once they have it. One example is working with TSA to open lanes ahead of demand rather than in response to it. The impact is dramatic, by the way: we saw a 20% increase in throughput at a New York airport. Another is a modest shift in counter allocations between airlines to better match the check-in profile by flight, improving processing times by 18%. These may not be glamorous AI use cases, but they are exactly the kind of thing that changes how a shift runs and directly affects passengers’ experience and perception of your airport.

Resource planning

The resource planning conversation sits alongside this, as airports are sitting on significant data and starting to ask harder questions about what they should be drawing from it. Predicting a gate conflict and preventing a late change 50 minutes earlier than you would today sounds modest. The operational knock-on if you don’t is not.

The same logic applies to baggage reclaim belts. Updating allocations to match the live flight schedule, and forecasting bags per flight rather than running on a plan made hours earlier, is a small shift in process with a disproportionate impact on the ground.

Data sharing

Data sharing sits underneath all of it. The appetite is genuine, but whoever holds the data knows they hold an advantage, and that changes the dynamic in the room quickly. The pragmatic instinct in most conversations is to start narrow, prove value, and build trust before trying to solve everything at once.

Pilots Are Easy. Production Is the Test.

Governance and procurement came up everywhere. AI is appearing on both sides of RFPs now, which is interesting in itself, but the path from pilot to production is still where momentum dies. Someone said it simply and it landed: pilots are easy because accountability is still fuzzy. Everyone laughed. Nobody disagreed.

Yes, there is always going to be a technical evaluation, but the airports making progress are treating this as an operational problem. They choose narrow use cases, set real accountability, and have a clear answer to who does what when the system flags something that needs attention. Procurement rules, legal review, and open-records obligations aren’t going away, and AI projects that can’t survive them won’t make it to production regardless of how well the pilot went.

Where Veovo Sits

This is familiar ground for Veovo, and we’re not standing on the sidelines of it. We build intelligence into the operational core: curbside monitoring, people flow understanding, lane planning, and gate and belt resource optimization. These get judged fast, in live environments, on the days when nothing goes to plan. We have real use cases with real outcomes, and we’re happy and proud to talk about them.

The questions being asked out there are the right ones. We’ve been building answers to them for a while.