navetta

QOI-78 — Proof of Value Proposal

CORTEX · HIA

Four sensors. One intelligence.

0.9%

Miss Rate

870

Simulated PAX

8 min

Prediction Lead

The Problem

02

Your four systems are powerful.
Together, they're blind.

54.3M PAX/year, 74% in transit. 64% say shorter queues are their #1 priority. Passengers spend only 4% of airport time in active processing — 25% just waiting. A 30-minute delay turns 40% of connections into missed flights.

40.7M transit PAX/year depend on a 45-min connection window. Four sensor systems generate data — but each speaks its own language, to its own dashboard.

“Fragmented data sources and lack of unified insights… limiting visibility into real-time passenger flows.”

Alessandro Milanes — MATAR, QOI-78 Webinar

◎

Cisco WiFi

Device location via access points. Zone-level heatmaps.

No journey tracking. 5-min lag.

◎

XOVIS Sensors

Stereo cameras. Queue length, wait times.

Point-only. No upstream prediction.

◎

SITA AODB

Flight schedules, gate changes, delays.

No link to PAX flow. Delay lands, nobody acts.

◎

Ipsotek VI-Suite

Video analytics. Anomaly, crowd density.

Reactive. 46% of delays are cascades.

The Solution

03

We don't add sensors.
We connect the ones you have.

Unify. Sensor fusion with adaptive weighting — signals are weighted by reliability, not just averaged. Four streams become one operational picture.
Predict. Predictive models trained on behavioral patterns. Not rules. Learning. Congestion detected 8–12 minutes before it forms.
Act. HIAS generates context-aware, trilingual PA messages. Not templates — AI-composed, situation-specific. One recommendation to the Duty Manager.
Learn. Reinforcement learning loop: every action's outcome feeds the next decision. The system improves with every flight. No manual tuning.

>70%

WiFi Penetration

±5%

XOVIS Accuracy

<30s

End-to-End Latency

What CORTEX delivers

Single dashboard for all 4 sensor streams
APGAR score — one number for airport health (0–10)
Density heatmap on HIA's real floor plan, all 5 floors
Predictive nudging — personal PA, buggy dispatch, staff escort
Gate hold logic — intelligent hold decisions when passengers are recoverable

Proprietary algorithms derived from battle-tested systems in regulated environments.

Live demo: hia.navetta.net

Architecture

04

Inspired by the human brain.

1

Synapse

Reads 4 data sources continuously. Protocol adapters for WiFi, XOVIS, SITA, Ipsotek.

2

Aura

Normalises and validates signals. Adds noise models for sensor imprecision.

3

Thalamus

Fuses inputs. Predictive coding separates real anomalies from noise. Dynamic gating.

4

Cortex

Predicts congestion. Recommends actions. Escalation chain from PA to gate hold.

5

Arena

Simulates outcome before committing. Behavioral profiles, queuing dynamics, full airport geometry.

Each layer is modular. Adding a new data source takes hours, not months. Replacing a sensor vendor requires changing one adapter — everything downstream stays identical.

Passengers with distinct behavioral profiles — Business, Tourist, Family, Elderly, PRM — move through HIA's complete Visioglobe geometry — every corridor, every floor, every gate. Real geometry, not a simplified diagram.

THALAMUS uses predictive coding — a technique borrowed from computational neuroscience — to distinguish genuine anomalies from sensor noise. Only surprises reach CORTEX.

SITA AODB adapter built and validated. Ready for live connection during POC kickoff.

Already Built

05

Not a mockup. Not a concept.
It runs today.

Live Dashboard

CORTEX 0.2 with sensor fusion, APGAR scoring, and density heatmap. Multi-flight simulation with aircraft, passengers, per-concourse transfer security. Real-time operational picture on HIA's actual map.

hia.navetta.net/cortex2.html

Predictive Intelligence

CORTEX doesn't just monitor — it anticipates. When congestion is forming, the system acts: personal PA, buggy dispatch, staff escort, gate hold. Each intervention is measured. Each outcome feeds the next decision. The system learns from every flight.

APGAR 0–10 — one score for airport health

AI-Powered Response

HIAS generates context-aware announcements in three languages — not from templates, but AI-composed for each situation. Personal PA for wandering passengers. Buggy dispatch for those too far from gate. Staff escort when nudges aren't enough. Every action targeted. Every result tracked.

On-Premise AI Engine

Our AI runs inside your walls. No cloud calls. No external APIs. No data crosses the perimeter. Proprietary ML models trained on airport behavioral patterns. Reinforcement learning that improves with every flight. Built for environments where security is not optional — it's the law.

Live now at hia.navetta.net — try it.

CORTEX in Action

06

In our simulation, CORTEX intervened 65 times. Here are five.

1

Passport control desks opened 40 min early

SITA: 3 flights landing in 20-min window, 280 PAX. WiFi: arrivals density rising. XOVIS: passport queue at 8 min, trend increasing.

CORTEX predicts queue will exceed 25 min. Recommends opening 3 additional desks now, 40 min before the wave hits.

The problem is solved before it exists.

2

Dedicated security lane for a delayed flight

SITA: QR572 from Delhi, 30 min late. DB: 42 transit booked on connections departing <50 min. XOVIS: transfer security concourse B already at 14 min.

42 transit in standard queue = missed connections. CORTEX organizes a dedicated security lane for QR572 transit. Staff reassigned from the least loaded lane.

The airport reorganizes around a single delayed flight.

3

Multilingual PA from flight demographics

WiFi: abnormal device concentration in duty free, 2× the average. SITA: QR178 to Malpensa boarding in 15 min. DB: 50% Chinese, 5% Italian among transit.

HIAS generates PA in English, Arabic, and Mandarin — only in the duty free zone. Gate, estimated walk time, urgency.

The right language, the right place, the right moment. Automatically.

4

Buggy fleet dispatched before the gap shows

SITA: QR302 boarding in 12 min. WiFi: device density at gate well below expected for a full flight. XOVIS: flow toward gate active but slow. Ipsotek: slow movement pattern in concourse corridor.

CORTEX estimates passengers still scattered in distant zones. Dispatches 3 buggies to lounge, food court, and concourse corridor.

Doesn't wait for a no-show. Goes and gets them.

5

Gate hold — four sensors, one decision

SITA: QR302 ready for pushback. WiFi: device density at gate still below manifested PAX. XOVIS: residual flow toward gate still active. Ipsotek: corridors to gate completely clear.

CORTEX cross-references all four. Flow is active, corridors are clear, passengers can make it in 5 min. Recommends gate hold +5 min. Dispatches buggy to last high-density zone.

The boldest decision: delay a flight. But data-driven, not instinct. Positive ROI.

Continuous Learning

07

The system that gets smarter
with every flight.

CORTEX uses Machine Learning to refine its predictions continuously. Reinforcement Learning — a branch of ML where the system learns from the outcomes of its own actions — drives the decision engine. Every intervention is measured: did the PA work? Did the buggy arrive in time? Did the gate hold save connections?

Three data sources feed the learning loop:

Historical data — existing airport records, past flight patterns, seasonal trends. The baseline.
Simulator outcomes — thousands of scenarios tested in the digital twin. Every new strategy is validated before it touches reality.
Live operations — real-time results from actual interventions. The ground truth that calibrates everything.

Simulator

Test new strategies
on digital twin

→

Live

Deploy gradually
measure results

→

Learn

Refine models
improve accuracy

←————— continuous feedback loop —————→

Simulator and reality converge

New routes, new devices, new solutions — tested in the simulator first. Always.
Gradual rollout. Every change is validated in simulation, then deployed live with monitoring.
The gap closes. As the system learns from real operations, simulator predictions become increasingly accurate.
Continuous refinement until strategies reach operational excellence. Not a one-time calibration — a permanent evolution.

Evaluation Match

08

Your 5 criteria. Our 5 answers.

MATAR Criterion	CORTEX·HIA Response	Status
1. Baseline & Accuracy	WiFi >70% penetration, XOVIS ±5% accuracy. Measured, not projected.	Target defined
2. Dashboard & Visualizations	Live dashboard running at hia.navetta.net. APGAR score, density heatmap, flight timeline.	Live today
3. Journey Time & Congestion	Check-in to gate tracked per PAX. Congestion predicted 8–12 min early.	Simulated & validated
4. Integration & Scalability	5-layer modular architecture. 2 zones to full airport = configuration, not development.	Architecture proven
5. Simulation Capability	Full-airport simulator on HIA's real geometry. Running today.	Running today

Bonus: APGAR airport vital signs — a single score (0–10) for operational health. Predictive nudging chain validated in simulation (PA → buggy → escort → gate hold).

Why Us

09

Specialists, not generalists.

In complex missions, precision teams outperform armies.

The POC Team

Cesare Rozzisi, Founder & CEO — System architect. HIA integration lead. On-site during critical phases (kickoff, calibration, delivery, validation). Available 24/7 throughout the POC.
ML Engineer — Sensor fusion, model training, simulator convergence. On-site during integration and calibration.
Integration Engineer — SYNAPSE adapters, dashboard, PA integration. On-site during API integration and testing.

Permanent Qatar base ensures continuity. Team rotation without operational interruption.

Track record — 8 PA clients + Regione Sicilia: KFlow, GPTPA, KPT — all production systems in Italy's most regulated environments.

Zero tolerance for data breaches. Zero tolerance for vulnerabilities. Security is the foundation every line of code is built on. Engineered and proven for the Italian government.

Why specialists win

The architect IS the CEO. Zero handoffs. You talk to who builds it.
AI on-premise. Our LLMs. Our code. No data leaves the perimeter. Not to the US. Not to China. Your data stays in Qatar, inside your walls.
Purpose-built AI tools. Our ML pipelines, our reinforcement learning models, our inference engines — built and controlled by us.
$40K co-invested. Skin in the game. Self-funded, profitable.
10-year development partnership. API-only integration. Core codebase never exposed.
Post-POC: Doha entity. JV with Qatari holding (MOU signed). QRDI-guided setup.

POC Plan

10

We don't consult. We embed.

Full immersion. Real problems. Engineered solutions.

1

Kickoff — Doha

Team on-site from day one. Living the operation, understanding pain points firsthand. SYNAPSE connected to SITA + WiFi.

2

Calibration

THALAMUS calibrated on HIA historical data. APGAR baseline established. Noise models tuned.

3

Training

CORTEX trained on live patterns. Thresholds tuned with HIA Duty Managers. First predictions.

4

Dashboard Live

Operational dashboard deployed. Trilingual PA connected. First recommendations to staff.

5

Full Integration

4/4 sensors live (+ XOVIS, Ipsotek). Virtual Twin fully calibrated on real HIA geometry.

6

Validation

30-day validation. 3 KPIs: prediction ≥8 min, alert-to-action ≤3 min, false positive ≤15%.

Budget — 6 Months

QRDI Grant $100,000

Navetta Co-Investment $40,000

Total $140,000

Our method: Our clients become our best partners. We immerse in their operations, absorb their pain points, and don't leave until the problems are solved.

We've done it in Italian Public Administration — environments where “it can't be done” is the default answer. We do it anyway.

Real problems. Simple, ingenious solutions. The best reward is the look on a client's face when something that seemed impossible just… works.

Post-POC: SaaS license model. $8–12K/month per airport zone. Revenue stays in Qatar through local JV entity.

Why This Works

Three reasons this works.

I

It runs.

Not a prototype. A working system on HIA's real geometry. The demo is live. See for yourself.

II

Zero hardware.

Software connects your 4 existing sensors. POC runs inside HIA data centre. No construction. No procurement delays. No capex.

III

A dedicated team. On the ground.

Not a remote consultancy. Not a vendor who flies in for workshops. Architect, ML engineer, integration engineer — on-site during every critical phase. Permanent Qatar base ensures continuity. Team rotation without operational interruption.

We use our own AI to write code. On-premise. Our code, from the first line, never leaves our machines.
For CORTEX·HIA, we guarantee the same: all code, all data, never leave the airport.
We already do this — with proven success — for the Italian government.