The Questions We Actually Get Asked

Yes, and reusing the existing fleet is the standard deployment model. Any camera that supports ONVIF or streams RTSP with H.264 or H.265 on a static IP can be connected; each module then sets the resolution it needs, and a site survey confirms it. Connection is not plug-and-play: a qualified network engineer configures each camera and router.

Yes. Customers retain 100% ownership and control of their data. IREX neither owns nor accesses customer video, events, logs, alarms, watchlists or floor plans, and there is no IREX-side data plane for customer media. That is an architectural fact rather than a contractual promise.

Yes, and that is the primary model. The platform installs as a private cloud on your own infrastructure: on-premises, private cloud, or fully air-gapped with no internet dependency at runtime. An IREX-hosted option exists for smaller deployments, and it is an opt-in convenience rather than a requirement.

By constraining what the platform can do, not by policy alone. It identifies only pre-registered people, meaning wanted suspects, missing persons and trafficking victims, and cannot recognize or track random individuals in public spaces. Every facial-recognition search requires a Case ID referencing a lawful basis such as a court order.

Every high-risk action is bound to a mandatory Case ID and written to a non-erasable, tamper-evident audit trail, with human review, daily signed log exports, and read access designed for supervisors and independent oversight bodies. Compliance auditing runs over analytics usage itself, so misuse surfaces as data.

No. Detections are signals for a person to verify, and no response runs autonomously.

The face recognition engine is trained on a proprietary 40-million-image dataset optimized for real-world CCTV and was ranked #1 in accuracy among US companies, top 10 globally of 268 algorithms, in its March 2021 submission to the NIST Face Recognition Vendor Test, on the FRVT 1:N Border and Kiosk datasets. IREX does not publish per-module accuracy figures, because a number from another city’s cameras tells you little about yours: accuracy is benchmarked on your own feeds during a pilot and the measured numbers go into the contract.

IREX publishes none, and would rather say that than invent one. What exists is the measurement machinery: per-deployment false-positive rates, accuracy trends and alert frequency tracked through dashboards, and pilots structured around two or three measurable criteria agreed in writing before they start. Where systematic bias exists, policy is to disclose it and its estimated impact.

The platform is GDPR- and CCPA-ready, and aligns with the EU AI Act, the NIST AI Risk Management Framework, and the FBI CJIS Security Policy. Ready is not the same as compliant: IREX is user-hosted, so full compliance is established per deployment, and each instance is validated separately. Encryption, role-based access and daily signed audit exports are built in. IREX supports all 13 CJIS policy areas, which is an alignment and compliance-support claim rather than a certification.

Neither, as IrexAI. The information-security management system is aligned with ISO/IEC 27001:2022 and 27002:2022 and holds no certification of its own; certification is a stated objective. For IREX-managed instances, the colocation layer carries the hosting provider’s own ISO/IEC 27001:2022 certificate, which is on file.

No. The commercial US SaaS runs in the IREX private cloud on bare-metal colocation in Dallas and Plano, Texas, which is ISO 27001 certified but is not a FedRAMP-authorized cloud. Where a program requires FedRAMP, a dedicated instance on an authorized cloud can be scoped and pursued per project.

For customer-hosted deployments, entirely on your own infrastructure in your own jurisdiction: on-premises deployment is the data-residency mechanism. For IREX-managed instances, in the IREX private cloud on colocation in Dallas and Plano, Texas, with backups inside the same managed environment. EU and EEA data subjects are served from an in-region instance rather than the US one.

A single instance is proven in production at 100,000+ cameras, all running real-time analytics with 30-day instant search, and the architecture scales linearly to 1,000,000 cameras, 100,000 users and a one-year retention period without architectural change. IREX manages 250,000+ cameras across 8 countries today.

The platform is licensed per camera and Sover per user, but we do not publish a price list and we do not quote a per-camera figure in isolation, because it is not the number your budget turns on. We quote the whole system: license, support, monitoring and services, mapped to the budget categories you hold, with a three-year total cost of ownership against your current stack.

Yes, and we prefer it. A typical pilot covers one site and three to five use cases over six to twelve weeks, with two or three measurable success criteria agreed in writing before it starts. Scope and cost are agreed case by case.

Six to twelve months for the IREX Ethical AI Platform and three to nine months for Sover, from first qualified conversation to steady-state operations. The dominant variable is the procurement route: sole source runs two to four times faster than an open tender.

Standard Support comes with the license: online ticketing, business-day coverage, 24×5 weekday response, and a four-hour initial response on an urgent production issue. Premier Support adds 24×7 deployment health monitoring, a one-hour urgent response, weekend escalation, senior-engineer access and non-production licenses. The availability SLA is 99.5% monthly with service credits.

You own and retain all right, title and interest in it throughout; IREX holds only a limited license to process it in order to provide the products. Either party may terminate for cause on 30 days’ notice for uncured material breach, and you may also terminate if hosted availability falls below 90% in a month for three consecutive months or any five months in a year, with a pro-rata refund of prepaid fees.