The FAA has expanded FAA AI tool SMART from Washington, DC to all 22 air traffic control centers. The system now reaches the full National Airspace System after an initial deployment that the FAA said improved routing and fuel use.
On September 21, SMART helped reroute 180 flights. The FAA said that initial deployment saved an average of 10 minutes per flight and almost 80 gallons of fuel per flight.
SMART’s route recommendations
SMART is a cloud-based system that combines weather patterns, flight paths, traffic flow and controller staffing metrics. It is designed to make scheduling recommendations, identify optimal flight routes and determine the best order of arrivals for aircraft.
That setup matters for passengers because the tool is steering traffic planning before controllers make the final call, not replacing the controller in the tower. A human team of FAA specialists based in DC validates SMART recommendations before they reach air traffic control facilities.
FAA rollout beyond Washington, DC
After the Washington, DC deployment, the FAA expanded SMART coverage to air route traffic control centers in Atlanta, Jacksonville and Miami. The nationwide step on November 10 puts the same system in use across all 22 centers, including traffic management tied to Dallas-Fort Worth International Airport.
The FAA also gave an example of SMART recommending a shorter and faster route for a cargo flight arriving at Dallas-Fort Worth International Airport. For travelers, that points to a system aimed at reducing wasted time and fuel across the network rather than making a visible change at any single airport.
Ed Bastian and union pushback
Ed Bastian and Robert Isom were among the airline executives who supported the rollout. The National Air Traffic Controllers Association said it was not consulted or involved in the design, testing or implementation of SMART.
The sharpest open question is how the system will perform when all 22 centers use it at once compared with the Washington, DC pilot. The FAA has put the tool into national use, but the broader result for delay reduction will come from how often controllers accept its recommendations.
