Physical security is the unloved cousin of cybersecurity — undercapitalized, dominated by aging vendors selling boxes of cameras and on-prem DVRs, structurally hostile to institutions that have neither budget nor staff to operate enterprise systems. Schools, churches, nonprofits, and small offices end up with security infrastructure that is either over-provisioned, under-monitored, or both. Arkc Technologies was built to address that gap with a different architecture: edge-AI inference on the camera, local decisioning on a small gateway, cloud only when the institution wants it.
The product family
Eight SKUs across two product lines, designed as a coherent system rather than a catalog. The MeshCam line is the sensor: six camera variants spanning indoor mini through outdoor vandal-rated, priced from $59 to $249, each carrying a Hailo-8L NPU capable of 26 TOPS of inference at 2.5 watts. The ML-Gateway line is the brain: two gateway variants providing local inference orchestration, 8–16 channels, optional cellular failover, and storage for 1–2TB of evidence retention.
The recommended deployment for the primary ICP — K–12 schools and mid-size nonprofits — pairs the ML-GW-200 Pro gateway with four to twelve MeshCam Outdoor or Indoor Pro cameras. A MeshBridge retrofit device at $149 lets institutions reuse existing camera investments during transition, framed explicitly as a sales bridge rather than a permanent category.
Site-local. Cloud-optional. Kid-safe by default. Edge-AI physical security built for institutions that the enterprise vendors gave up on.
The defensible differentiator
The primary defensible feature is cellular failover at the gateway. School networks fail. Power blips. Internet provisioning is unreliable in the buildings that need security infrastructure most. An ML-GW-200 with cellular failover continues operating — inference, recording, alerting — through outages that would render conventional systems blind. The technical implementation is straightforward; the institutional confidence it builds is not.
The Hailo-8L NPU at the camera edge is the second structural differentiator. Inference runs on-device, which means privacy decisions happen before frames ever transit the network. For school deployments — where federal student-privacy obligations are non-negotiable and the political cost of an incident is catastrophic — edge-only inference is the architectural answer to questions that cloud-based competitors cannot answer at all.
The architecture
Three layers, each with a single responsibility. L1 — Sense: the MeshCam edge devices capture and pre-process. L2 — Infer: the ML-Gateway aggregates, reasons, and decides locally. L3 — Govern: optional cloud sync surfaces an admin dashboard and cross-site analytics for institutions that opt in. The data path is sense → infer → govern, with the institutional ability to terminate the flow at any layer.
The go-to-market
K–12 schools, faith-based institutions, and nonprofits in the Newark and tri-state region are the first ICP. Procurement is sales-led with channel partners, not direct-to-school online ordering — institutional buyers want a human conversation and a written deployment plan, not a shopping cart. Pricing is presented as a three-year total-cost-of-ownership comparable to incumbent enterprise systems while delivering capabilities the incumbents do not offer.
Status
Hardware development. NDA-bound on the specific design of the Hailo-8L integration and the cellular-failover module. The eight-SKU pricing structure is locked. The retrofit MeshBridge device is in prototype. First institutional deployment scheduled for late 2026.