The Federal Aviation Administration has begun rolling out an AI-supported system designed to spot potential flight delays before aircraft leave the gate, marking a major shift toward more predictive air traffic planning in the U.S. national airspace.

Get the latest news straight to your inbox!

FAA rolls out AI tool to predict flight delays before takeoff

What the FAA is launching and where it starts

Published coverage and FAA materials describe the new capability as Strategic Management of Airspace, Routes and Trajectories, or SMART, a planning layer intended to help the agency anticipate schedule conflicts, capacity limits, and weather disruptions before they cascade into day-of-travel delays.

On September 21, 2026, the FAA began using SMART in a limited mode focused on the airspace around Washington, D.C., with expansion planned in stages to other parts of the country. Separate reporting about disruptions in the Northeast also referenced the agency beginning to test an AI-based computer system in the Washington area that is not connected to communications outages, reinforcing that early deployment is localized and not yet nationwide.

The system is positioned as decision support rather than automated control. Publicly available FAA descriptions emphasize that air traffic controllers remain responsible for separation and safety, while SMART produces forecasts and planning recommendations that can be reviewed and accepted or declined by FAA personnel.

How SMART predicts delays before departures

SMART is described as a cloud-based platform that builds a shared view of conditions across the National Airspace System before flights depart. It synthesizes many inputs at once, including airline schedules, weather, airport capacity, airspace conditions, and operational constraints.

The practical idea is to identify problems earlier, when there is still room to make changes that are less disruptive to travelers. Instead of waiting for congestion to build and then issuing tactical fixes, the tool is designed to surface where demand is expected to exceed capacity and where weather or constrained routes will create bottlenecks.

In FAA descriptions, SMART continuously updates forecasts as conditions evolve through the day, with outputs intended to support decisions on timing and routing choices. That includes highlighting potential conflicts and offering a more centralized visualization of how today’s schedule interacts with predicted capacity and constraints.

The larger modernization push behind the AI system

SMART is closely tied to Flow Management Data and Services, or FMDS, which the FAA describes as the future backbone for its traffic flow management modernization. FMDS is intended to replace the legacy Traffic Flow Management System and to project congestion hours in advance using flight plans, airline schedules, and real-time position updates.

The FAA has described FMDS as part of a broader effort to consolidate data that is currently spread across multiple tools, displays, and manual workflows. Public materials describe the goal as improving reliability and speed of traffic management automation, while simplifying how managers view and act on system constraints.

The FAA has also framed the combination of SMART and FMDS as enabling more proactive flow management by allowing planners to balance demand and capacity earlier, coordinate reroutes around severe weather more precisely, and improve collaborative decision-making with airspace users.

What this could mean for travelers and airline operations

For passengers, the immediate promise is not a new consumer-facing delay prediction app, but improved system-level planning that can reduce the chance that flights start the day behind schedule. FAA descriptions link the approach to fewer tarmac waits, fewer airborne holding patterns, and fewer diversions, especially during high-impact weather and congestion events.

The tool’s emphasis on pre-departure planning reflects how delays often compound: once the system is saturated, small disruptions can propagate through aircraft rotations, crew schedules, gate availability, and airport surface movement. A model that flags conflicts earlier could allow airlines and the FAA to adjust timing and routing before aircraft push back, which is typically less disruptive than managing a backlog after congestion has already formed.

SMART’s early rollout also arrives against a backdrop of ongoing reliability concerns in parts of the air traffic infrastructure. Recent disruption coverage in the Northeast highlighted how technical problems at key facilities can quickly affect multiple major airports, underscoring why predictive planning tools and modernization efforts are being pursued in parallel.

What to watch next as deployment expands

Travelers should expect the initial impact to be uneven because SMART is being introduced in stages, beginning with Washington-area airspace. The FAA has described the rollout as gradual, implying that measurable improvements, if they materialize, may first appear in the regions where the system is actively used and integrated into daily planning routines.

Another key question is how the tool’s recommendations translate into operational choices. Public descriptions emphasize that recommendations are reviewed by FAA staff and can be accepted or declined by local facility leadership, which suggests outcomes will depend not just on model performance but also on how the system is used in practice.

As SMART expands, observers will be watching whether earlier identification of demand-capacity mismatches leads to fewer broad traffic management initiatives, faster recovery after storms, and fewer knock-on delays later in the day. For now, the clearest verified detail is that the FAA has moved from announcing the concept to limited operational use in the Washington, D.C., region as of September 21, 2026.