← Aviation · 02

Ten-second clearance. Audit-grade rationale.

GRU runs 42 + 28 SITA gates. JFK NTO + iProov launched Feb 2026. The pitch leads with governance, not throughput.

Proven pattern

Biometric face-match plus risk-scoring for entry/exit at the border: government ID-document verification, face capture, watchlist check, risk-triage agent that routes low-risk passengers through eGate and high-risk passengers to secondary inspection.

Hard KPIs · operator-disclosed where possible
~10s
per-passenger clearance (GRU)
>99%
face-match success in TSA/DHS scenario testing (PCLOB)
238
US airports running CBP TVS

Buyer · Director of Operations + government partners (SAT/INM in Mexico, PF in Brazil, AFIP/Migraciones in Argentina, CBP/TSA in US).

Operator references

Provenance tiered explicitly. Audited > Operator-disclosed > Vendor-published. Treat all vendor-tier figures as directional.

GRU São Paulo (Brazil)
SITA biometric ABC eGates: 42 in Phase 1 (Dec 2025), 28 more in Phase 2 (Feb 2026), ~10-second clearance. Phase 2 includes a smart-security pilot with CT scanners and biometric boarding.
Operator-disclosed
Ezeiza Buenos Aires
Vision-Box biometric eGates in T2 since 2023.
Operator-disclosed
Santiago SCL (Nuevo Pudahuel / Vinci)
mONA facial-recognition mobile app for end-to-end checkpoints. Ikusi integrated security-and-operations control center for Terminal 2 (2022).
Operator-disclosed
JFK New Terminal One (US)
CBP Enhanced Passenger Processing with iProov biometrics for US-citizen arrivals (Feb 2026 launch). Part of $19B JFK transformation.
Operator-disclosed
US CBP TVS + TSA CAT-2
CBP TVS for entry/exit at 238 US airports. TSA CAT-2 facial-match devices at 30+ US airports. BigBear.ai supplies EPP biometric tech at JFK, LAX + 10 other US ports of entry.
Audited
PCLOB / GAO (US, governance)
TSA/DHS scenario testing showed >99% face-capture and match success in fall 2024. GAO has warned that real-world performance gaps remain and that demographic and operational factors still matter.
Audited
Why agents beat traditional ABC gates

Risk-triage logic is the wedge. Pure face-match is commoditized; the value is in the multi-signal scoring and the dynamic routing to secondary. The agent must produce a structured, auditable risk-score rationale that holds up under both internal review and PCLOB-equivalent oversight.

Why this is hard

Government-relations dependency is total. This is not a private-sector procurement. Privacy and civil-liberties oversight is active and increasing in the US; LATAM governance frameworks vary widely. Demographic-disaggregation evaluation is a hard technical problem and a hard political problem at the same time.

Mexico · LATAM specifics

Mexico's INM (immigration) and SAT (customs) have not publicly announced a biometric-eGate program at GRU/JFK scale; AICM (Mexico City) operational caps make it the natural high-value target. Cancún (ASUR) handles the highest tourist-passenger throughput in Mexico, the natural visitor-experience pilot surface.

Reference architecture

Each node maps to a regulatory anchor.

  1. 01
    Capture: government ID + face
  2. 02
    Match: biometric template comparison
  3. 03
    Risk score: multi-signal (watchlist, behavior, manifest)
  4. 04
    Routing decision: eGate or secondary
  5. 05
    Human-officer handoff with structured risk rationale
  6. 06
    Audit log + demographic-disaggregation reporting
Anti-positioning

Not "Frictionless surveillance." Surveillance-state visual vocabulary. Throughput as the lead.

But Governance and oversight first. Throughput second. The oversight loop is a structural feature, not a footnote.

What didn't work initially

Early deployments fail when demographic-disaggregation evaluation is left to launch+1. The PCLOB and GAO playbook is explicit: test before deploy, report by demographic, and treat any gap as a hold-the-launch event.

A working session, not a sales call.

Two hours with a partner. Your incumbent stack, your data posture, and the regulatory surface against a sovereign reference architecture for aviation.

Request a working session →

A working session, not a sales call.

Two hours with a partner. We map your AI spend, data exposure, and governance posture against a sovereign reference architecture. You leave with a memo. We leave with a decision.

By invitation.