← Aviation · 03

The cameras are already there.

Port Authority + NYU at JFK T4: 15% density reduction. AICM curbside is the natural Mexican anchor.

Proven pattern

Computer vision on existing CCTV detecting queue formation, dwell time, curbside congestion, rideshare/taxi/shuttle mix; feeding a decision-support agent that recommends interventions (additional staff, lane reassignment, signage updates). The lowest-capex AI deployment in the airport dossier.

Hard KPIs · operator-disclosed where possible
15%
traffic-density reduction (PANYNJ/JFK T4)
57→96%
shuttle-bus detection accuracy on existing cameras
90 days
pilot-to-measurement window for one terminal

Buyer · COO + Airport Operations Center lead + landside-operations head. Easier to procure than airside CV; fewer cross-stakeholder dependencies.

Operator references

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

Port Authority of NY/NJ + NYU pilot
Existing cameras for queue detection and curbside monitoring. ~90% accuracy for passenger cars; shuttle-bus detection improved from 57% to 96%. JFK T4 case: 15% reduction in traffic density with shorter dwell times after the remote for-hire vehicle process was introduced.
Operator-disclosed
Bogotá El Dorado (OPAIN) · DoraBot
Botmaker-built, GPT-integrated WhatsApp virtual assistant (Nov 2023). 24/7 flight info, queue waits, check-in guidance. Cybersecurity AI: Sofistic/Cuatroochenta + Darktrace 24/7 SOC; <2-hour incident response.
Operator-disclosed
Lima Jorge Chávez T2 (LAP/Fraport)
Living Map digital wayfinding (2025) plus location-based push notifications; biometric infrastructure; remote ATC; airport-wide capacity target 40M PAX.
Operator-disclosed
GRU São Paulo
ARINC VeriPax passenger-reconciliation 15–20% flow improvement. Vendor-aligned, directional.
Vendor-published
Why agents beat passive monitoring

The win is the recommendation layer, not the detection layer. Detection is increasingly commoditized; the agent's value is in turning the signal into staffed action.

Why this is hard

Compared to airside, this is the easier sale, but procurement still runs through public-sector cycles. Coordination with municipal traffic and rideshare/taxi/shuttle operators is non-trivial.

Mexico · LATAM specifics

AICM's terminal-area curbside chaos is well-documented; the natural anchor pilot. AIFA's policy-driven traffic shifts make landside dynamic-routing even more valuable. Guadalajara (GAP) and Monterrey (OMA) are clean secondary-hub candidates.

Reference architecture

Each node maps to a regulatory anchor.

  1. 01
    Existing CCTV: no new hardware
  2. 02
    CV detection: passenger cars, shuttles, rideshare, taxis
  3. 03
    State representation: queue length, dwell, congestion, mix
  4. 04
    Recommendation agent: intervention with confidence + ETA
  5. 05
    Ops-center dashboard with named operator approval
  6. 06
    Staff dispatch / signage update / lane reassignment
  7. 07
    Measurement loop: dwell, density, complaint rate
Anti-positioning

Not A "smart city" cliché. A multi-year transformation. A new-capex pitch.

But A 90-day proof at one terminal with one clear KPI, then expand.

What didn't work initially

Early deployments fail when the recommendation layer is rolled out without a named ops-center approver. The signal is good; nobody owns the action. The fix: every recommendation has an owner, an SLA, and a logged outcome.

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.