Part 2: SMART and the Four-Phase Framework

The FAA's Strategic Management of Airspace Routing Trajectories program, as analyzed through the four-phase automation framework developed by Leeham News & Analysis contributor Vincent Bianco, occupies Phase 1 of a deployment architecture that ranges from passive data aggregation to full NAS infrastructure integration with real-time routing authority. SMART extends conflict prediction from a 15-minute window to two hours by ingesting flight schedule, weather, and airspace state data into a 4D modeling engine, producing scheduling adjustment recommendations for controllers who retain full decision authority. The system does not issue instructions, override clearances, or contact flight crews directly — it feeds human controllers who remain the authoritative layer between the algorithm's output and actual aircraft separation. Bianco's framework uses ACAS X as the reference case for a more advanced Phase 3 deployment: a system with certified advisory authority that pilots are required to follow during conflict events, bounded by scope, time, and equipage requirements developed over 15 years of rigorous verification at MIT Lincoln Laboratory before TSO approval. SMART's architectural placement at Phase 1 is, by the framework's own logic, the appropriate starting point for a system of this scope.

The substantive concern Bianco raises is not the architecture but the procurement velocity. The FAA announced a competitive vendor selection among Palantir Technologies, Thales SA, and Air Space Intelligence with first operational deployment targeted for September 2026 — approximately five months from public announcement to live NAS deployment. That timeline is anomalous against every relevant federal aviation precedent: NextGen consumed $7.5 billion over 15 years and reached 16 percent completion; ACAS X required a 15-year arc from research origin to operational certification. The compression is attributed explicitly to political rather than technical drivers. FAA Administrator Bryan Bedford, confirmed in July 2025, has made SMART a personal agency priority. The Department of Government Efficiency applied external pressure through facility visits and public commitments to rapid upgrades. The March 2026 LaGuardia collision created public urgency that the political layer is channeling toward visible near-term deployment rather than the structural institutional reforms the NTSB identified in its DCA collision analysis as the underlying pathology. For pilots and operators, the significance is not that political pressure produced a bad system — that remains unknown — but that the procurement velocity structurally obscures whether verification discipline is being applied at all.

The vendor landscape provides partial signal on the verification question. Air Space Intelligence's Flyways AI system has been advising Alaska Airlines dispatchers on flight-path optimization since 2021, giving it a five-year deployment history at a Part 121 carrier with the closest existing analog to the kind of operational feedback loop that informs responsible federal deployment. Palantir's Foundry platform has existing FAA data infrastructure integration through prior sole-source contracts, which represents genuine technical depth in the agency's data environment but does not constitute verification of the AI advisory logic itself. Thales brings 85 years of ATM heritage and dominant European market position, though its U.S. deployment context is comparatively unfamiliar. None of the three vendors possesses a verification arc equivalent to Lincoln Laboratory's ACAS X development program. The article ends before completing the implications of that gap, but the framing makes the policy question explicit: whether the vendor selected in the compressed timeline will have performed — or will be required to perform — the verification work commensurate with deploying advisory software into live NAS infrastructure.

For Part 121 dispatchers, Part 135 operators, and flight departments coordinating complex routings through congested NAS corridors, SMART's two-hour conflict prediction horizon represents a meaningful operational capability if it performs as described. The current 15-minute ATC planning window creates downstream inefficiencies that cascade into fuel burn, schedule reliability, and contingency planning across the NAS. A validated advisory system with that extended horizon could materially improve pre-departure routing decisions and reduce inflight rerouting workload. However, the verification question matters operationally as well as institutionally: advisory systems that produce high false-positive rates or fail to account correctly for dynamic weather and traffic interactions can degrade controller trust and workload rather than reduce it, a failure mode documented in prior ATC automation initiatives. The distinction Bianco draws between verified and merely deployed advisory authority is not abstract — it is the practical difference between a tool that controllers integrate confidently into separation work and one they learn to filter or disregard.

The broader trend the analysis reflects is the increasing convergence of political timelines with aviation safety infrastructure decisions in a period of visible NAS stress. The DCA collision, the LaGuardia incident, controller staffing shortfalls, and the DOGE engagement with FAA operations have collectively created a policy environment in which aviation technology modernization is being evaluated on visibility and speed rather than exclusively on the verification standards the industry's safety record was built upon. SMART, as a Phase 1 advisory system with no direct aircraft control authority, carries lower immediate safety stakes than a Phase 3 or Phase 4 deployment would. But the procurement pattern it establishes — and whether the FAA's contracting requirements for this program include meaningful algorithmic verification standards — will set precedent for how Phase 2, 3, and 4 systems are eventually procured as automation ambitions in ATM continue to escalate.