Beginning September 21, 2026, the skies over the Washington, D.C., region entered a new phase of aviation technology.
For the first time, the Federal Aviation Administration began using a new artificial-intelligence-supported system designed to help anticipate air-traffic congestion before it becomes a problem.
The system is called SMART, short for Strategic Management of Airspace, Routes and Trajectories.
Its first operational deployment covers three of the region's busiest airports: Ronald Reagan Washington National Airport, Washington Dulles International Airport and Baltimore/Washington International Thurgood Marshall Airport.
The idea behind SMART is deceptively simple.
Instead of waiting for congestion, bad weather, limited airport capacity or other disruptions to create delays and then responding to them, the FAA wants to identify those problems earlier.
SMART brings together large quantities of information, analyzes how those conditions could interact and provides predictions and recommendations to aviation specialists.
The system is designed to look ahead.
Where will traffic become too heavy?
Which airports may become constrained?
How could weather affect aircraft flows?
Could changing departure times prevent a bottleneck?
Could another route reduce pressure on an already crowded section of airspace?
And if a problem appears likely several hours in advance, what can be done before thousands of passengers are affected?
Those are the kinds of questions SMART is intended to help answer.
But there is an equally important point about what the new technology does not do.
It does not fly airplanes.
It does not issue clearances on its own.
It does not replace air-traffic controllers.
Controllers remain responsible for safely separating aircraft and making operational decisions. The FAA describes SMART as a planning layer that works alongside existing systems rather than as an autonomous replacement for the people responsible for controlling aircraft.
That distinction is central to understanding what the FAA is actually changing.
The United States is not handing control of its airspace to an algorithm.
It is attempting to give human aviation professionals a much better forecast of what is about to happen.
Why Washington Comes First
The Washington region is an unusually complicated place to test a system like this.
Three major commercial airports operate within the same broad metropolitan airspace.
Reagan National sits close to downtown Washington.
Dulles is farther west in Northern Virginia.
Baltimore/Washington International is northeast of Washington near Baltimore.
Their operations overlap with other aviation activity, including military operations, restricted and special-use airspace, helicopter traffic and other constraints.
The FAA says the National Capital Region has particularly complex airspace restrictions. A Special Flight Rules Area extends 30 miles around Reagan National, with an inner 15-mile ring subject to even tighter operating requirements.
The FAA's Potomac Terminal Radar Approach Control, or TRACON, manages aircraft approaching and departing the Washington-area airports, including DCA, IAD and BWI. The agency says roughly 600 highly technical employees work at the facilities responsible for managing this traffic environment.
That makes Washington a useful laboratory.
If SMART can successfully predict and help manage congestion in an environment this complicated, the FAA will have valuable information before expanding the technology to other parts of the country.
The first phase is therefore deliberately limited.
Rather than immediately deploying the system across the entire National Airspace System, the FAA has begun with a roughly 90-day operational trial in the Washington region.
The goal is not simply to demonstrate that artificial intelligence can generate predictions.
The goal is to find out whether those predictions are useful enough, accurate enough and operationally practical enough to improve the way the aviation system works.
What SMART Actually Sees
The most important feature of SMART may be the amount of information it can combine.
The FAA says the platform centralizes approximately 200 data streams, including weather patterns, flight paths, traffic flows and controller staffing information.
The FAA's SMART fact sheet describes the system as creating a shared real-time picture of schedules, weather, airport capacity and airspace conditions. It is designed to identify constraints and conflicts earlier, while there is still time to address them.
That is fundamentally different from simply putting another screen in front of a controller.
The challenge in modern aviation is not necessarily a lack of data.
There is enormous amounts of data.
Airlines know their schedules.
Airports know runway and gate availability.
Weather services know where storms are developing.
The FAA has radar, surveillance, flight-plan and airspace information.
Controllers know what is happening in their sectors.
The difficulty is combining all of those pieces quickly enough to understand what they mean collectively.
A thunderstorm over one airport might force aircraft to divert.
Those diversions can increase traffic elsewhere.
The resulting congestion may cause holding.
Holding consumes fuel and occupies airspace.
That can create further delays.
Those delays can cause connecting passengers to miss flights.
Airlines may then need to reposition aircraft and crews.
A problem that began with a single weather cell can therefore spread through the network.
SMART is designed to recognize those interactions earlier.
From Reaction to Prediction
Traditional air-traffic management often requires people to respond to conditions as they develop.
That is unavoidable to some degree.
Weather changes.
Aircraft arrive late.
Runways become unavailable.