Turn any camera network into an intelligent, ethical security ecosystem.
AI now takes a goal, plans the steps, and runs the whole task — not just a single answer to a single question.
Cameras that understand plain language. New detectors appear from a sentence — no datasets, no training cycles.
The EU AI Act era: transparency, human oversight, and non-erasable audit trails become mandatory for high-risk AI.
Rising geopolitical conflict pushes nations to run critical AI on their own hardware, inside their own borders.
Specialized AI modules watching every camera, every second.
Video search and Ask IREX run multi-step searches from plain language.
StreamVLM™ turns a sentence into a working detector.
Transparency by Design — a Case ID on every high-risk action.
Your hardware, your data, national scale.
No rip-and-replace — an upgrade path, not a forklift.
IREX brings the cameras and sensors you already operate onto one secure platform, then puts what they see in the hands of the people who keep communities safe.
The networks you already operate
Aerial video from patrol and response UAVs
Live signals from across your sites
Field photos and video from any device
Faster, better-informed response
Safer transit, roads, and stations
Search and investigate with a Case ID
Coordinated, jurisdiction-wide oversight
StreamVLM™ turns a text prompt into a working detector — no datasets, no training, no waiting. Shipping in beta on selected instances; included in the standard IREX price.
What is a Vision-Language Model? An AI model that has learned to connect what it sees with the words people use to describe it — so it can be told what to look for in a sentence, instead of being trained on thousands of labelled examples first.
Search people, vehicles, and events across petabytes of video — by map, filter, or natural language (English, Spanish, Arabic). The investigation agent reads intent, cross-references watchlists, and runs the multi-step search through the IREX API.
Proprietary descriptor clustering and indexing searches through billions of objects — faces, person figures, and vehicles.
Natural-language queries and investigation agents — from the main IREX interface and from a Sover chat — same Case ID, same permissions, same audit log.
Lawful search, bound to a Case ID
Plate, make, model and color search
Intrusion, loitering, line-crossing
People on rail track, unattended luggage
Density, flow and crowding alerts
Across non-overlapping cameras
Flow, violations and corridor tracking
A new detector from a single sentence
“Asian, glasses, moustache”
Early visual detection of fire and smoke
Obstruction, defocus and outage
Route reconstruction across cameras
Before an operator can search, watch, or export anything, the platform demands a Case ID: a reference to the lawful grounds for that action — a case file, court order, or missing-person report. Who acted, what they did, when, and why is written to one consolidated, non-erasable Logbook, exported daily as digitally signed files that supervisors and independent oversight bodies can audit.
Every action — searches, database edits, alert configuration, exports, permission changes — lands in a searchable audit log that cannot be erased.
Private cloud on open-source components: TLS 1.3 end to end, WAF/IDS/IPS in depth, JWT with rotating keys, MFA, and penetration testing every major release.
Pre-identified people only — wanted suspects, missing children, trafficking victims. Biometric database size is capped; random individuals cannot be recognized or tracked.
Hierarchical role-based access: cameras, databases, modules, alerts, and results appear only for the roles that need them. Configuration is top-role only, and logged in full.
Balanced training data and continuous tuning on a 40-million-image dataset optimized for CCTV. Where systematic bias exists, IREX discloses it and its estimated impact.
User guidelines, audit processes, and compliance checklists built with outside experts — designed to support FBI CJIS, GDPR, and the EU AI Act.
The cameras, the servers, and the people who use them all sit inside one network you control — nothing leaves it unless you decide.
Figures reflect IREX platform deployments and can be confirmed under NDA. Outcomes vary by site, configuration, and agency processes.
National Police and multi-city command centers
London Underground safety and critical infrastructure
Southern California law enforcement, multi-jurisdiction
Dublin Airport: watchlists, unattended items, weapons
Nationwide multi-camera traffic enforcement
National smart-city program, full AI module catalog
Premier League stadium security, banned-fan watchlists
Retail loss prevention, repeat-offender alerts
Seven stage-gated phases. Each gate has named owners and a written output before the next one starts.
Qualification against the live demo zone; output is an opportunity brief.
IREXSites, cameras, networks, integrations and sovereignty posture mapped.
JointTime-boxed 4–8 weeks on real cameras, pre-agreed success criteria.
IREXProduction sizing, deployment model, and the commercial package.
JointContracting, hardware, site access and the delivery team stood up.
PartnerPrivate cloud built, cameras onboarded, analytics and integrations live.
JointOperator training, hypercare, then steady-state run and support.
JointIREX owns the platform & cloud (L3/L4); the certified integration partner owns site, network, onboarding & first-line support (L1/L2). Governance: weekly delivery standup, bi-weekly steering committee, monthly executive review.
Co-Founder & CEO
President, CTO & Co-Founder
Advisor, Ethics & Compliance
Advisor, Investments
Advisor, Partnerships
Director of Operations & Deployment
Business Development
Sales Operations
Business Development, LATAM
Public Sector Partnerships
Leading technology companies extend the platform — and mission-driven organizations put it to work protecting people.