Justin Keene is product manager at Veovo, specializing in resource planning and management. He began his career as a programmer and technical consultant, working hands-on with airports to implement RMS platforms. That grounding shaped his view of how plans behave under pressure. Today, he leads the rollout of a new generation of AI-assisted resource planning tools
A gate or stand plan always looks calm at first glance. Pucks align. Buffers sit neatly between turns. On a Gantt chart, it is reassuring.
On the apron, it rarely survives the morning.
Arrivals compress and a wide-body holds the stand longer than expected. What looked settled at dawn is already under strain by mid-morning. At slot-constrained airports, the tension sharpens. The runway may be full, but slots only govern when aircraft arrive and depart. They say nothing about where those aircraft sit once they touch down.
That is where the plan starts to fray. Stand conflicts appear and pushbacks begin to slip.
Over the years, I have watched planners wrestle with this across three horizons. The seasonal plan sets the framework months ahead. Rolling planning, six to 12 weeks out, absorbs the churn of charters, reschedules and works. Then there is the day of operations, when weather and delays can undo even the neatest plan.
AI now has a role in all three, for different reasons.
Bridging the gap between slots and stands
Slot files were designed for runways, not aprons. They assume average behavior. The ground does not.
An A321 with a fast turn can be boxed in by a wide-body that needs twice the time. When arrivals tighten, the problems compound.
I saw this clearly at a hub running close to 60 movements an hour. During the morning wave, gate conflict rates hovered around 3%. That meant nearly two aircraft an hour competing for the same gate. Each clash added roughly 10 minutes of primary delay, with knock-on effects for connections and crews.
With AI in the loop, the slot view becomes airport-aware. Forecasts of real off-block times, rather than averages, flag overlaps well before they happen. The number of conflicts is halved and multiple turns are saved in a single rush period.
Seasonal gate planning: starting strong
Seasonal plans lock in half a year at a time. They are revisited, until the first ripple of operational reality a few weeks out.
The risks sit quietly. Buffers are applied uniformly and some gate allocations fail under pressure. Curfews drift without their operational impact being fully tested.
AI strengthens the plan at the outset. It learns how turns actually behave by fleet and by time of day, making instinct visible.
At one airport averaging 40 tow moves a day, that insight reduced demand by around 20%. Eight fewer moves daily meant less tug time, less fuel burned and smoother first-wave performance. Small changes, multiplied across a season, made a measurable difference.
The rolling plan: staying ahead
Schedules never sit still. Charters appear and aircraft swap, while stands close for works. I have seen planners spend evenings in spreadsheets, nudging flights and hoping the structure would hold.
Here, the work changes character. Instead of rebuilding the plan, the system re-optimizes around what has shifted. It protects connections and airline agreements without breaking operational rules.
Planners spend less time rebuilding and more time deciding. They review a short list of viable swaps, ranked by impact.
Day of operations: recovering faster
This is where time matters most.
A wave lands early, and a long-haul holds the stand. Too often, issues like this surfaced only when the tug was already on its way.
AI pushes the warning earlier. Models trained on weather and upstream delays, grounded in historical turn behavior, flag likely clashes hours ahead. Supervisors see credible alternatives, with the logic laid out.
One example stays with me. Two A320s were planned on the same gate, separated by a generic 30-minute buffer. The second fed a bank of long-haul transfers. The model predicted the first would overrun by 42 minutes and suggested a different stand for the second, hours out.
Ground crews adjusted, passengers made their connections and the airline avoided the far greater cost of missed bags, rebooked itineraries and compensation payouts. The issue never became urgent.
Why AI in gate management matters
The value of AI here is precision and lead time.It replaces averages with forecasts that reflect how operations actually behave. Gates are assessed on when they will be free, not when they should be free. Most conflicts are identified early enough to be handled routinely. In airport operations, that often makes the difference between a plan that holds and one that unravels.
Connecting the horizons
For years, airports have worked with separate tools: one for slots, another for the seasonal plan, and a third for the day itself. Each solved part of the problem, but none joined it up.
AI shifts that balance. It connects the horizons into one flow and pulls in data that planners never had at their fingertips before: stand use, real turn behavior, upstream delays, weather impacts. Suddenly, what used to be hidden becomes visible, and what used to be reactive becomes manageable.
I think of it as giving planners room to plan rather than patch. One connected view, holding from the first draft of the season through to the last departure of the day.
In airport operations, that kind of change is usually the only kind that sticks.