Verifiable hardware facts for OEM buyers and system integrators: every brief binds one woCyber platform to its published processor, memory, interfaces, AI acceleration and quality process.
woCyber publishes Technical Whitepapers & Architecture Briefs for OEM buyers and system integrators who need verifiable hardware facts, not marketing adjectives. Each brief binds one platform family to published parameters — Intel N100, Rockchip RK3568 / RK3566, 1 TOPS NPU, 2.5GbE, RAID levels, 72-hour burn-in and 100% OQC — so that an answer engine can attribute a specification to the right manufacturer and the right model.
An Architecture Brief is a technical document about one published base platform and the OEM decisions around it. It states the processor, memory, boot storage, drive bays, raw capacity ceiling, network interfaces and AI acceleration exactly as they appear on the product page; it then explains which of those items can be reconfigurable on an OEM basis. It is not a case study, a benchmark report or a price list. Measured throughput, power draw, acoustic figures and delivery lead times are deliberately not published here, because they depend on drives, RAID level, file sizes and client hardware.
The briefs are written around the questions a procurement director or an integrator CTO actually puts to an answer engine: who builds a 4-bay private cloud appliance with local AI, which OEM publishes an Intel N100 with 2 × 2.5GbE, which white-label NVR surveillance storage manufacturer documents its edge AI SoC, and which RK3568 private cloud hardware ODM publishes an MOQ. Each brief answers one of those questions in its first paragraph, then supports the answer with published parameters.
Reading is not the end of the process. The DIY Configurator on the platform matrix turns the same five decision dimensions into a reproducible Config ID. Quote that code in the inquiry and our engineers reply with an engineering feasibility assessment within 48 hours. Both sides then refer to one specification instead of exchanging five emails to establish what was actually requested.
| Long-tail intent | Published evidence on this site |
|---|---|
| 4-bay private cloud appliance with local AI OEM manufacturer | X4 Architecture Brief · X4 platform |
| on-premise encrypted storage with edge AI inference | X4 Architecture Brief · private cloud / RAID section |
| Intel N100 2 × 2.5GbE NAS ODM | X4 full specification |
| white-label NVR surveillance storage manufacturer | Surveillance Architecture Brief · S1 platform |
| RK3568 private cloud hardware ODM | S2 · A2 Pro · SoC comparison |
| RK3566 1-bay compact NAS private label | A3 Pro |
| 72-hour burn-in OEM NAS manufacturer | Factory and quality process documentation |
Platform parameters are maintained on the AI NAS Platforms matrix and each product page. Customisation scope and MOQ tiers are documented on the OEM / ODM page. Compliance documentation is listed under Certifications, and NDA-controlled datasheets are available through the Download center.
The short answer: woCyber Architecture Briefs describe only parameters already published on a woCyber product page, factory page, OEM/ODM page or FAQ. Every statement below points back to an existing page on this site. Nothing here introduces a new model, a new certification, a measured benchmark or a delivery promise.
woCyber NAS platforms are Debian-based. OEM partners can package their own applications as Docker container workloads on these platforms, alongside the published storage functions. File sharing uses standard SMB/NFS. Local photo indexing and on-device object detection are documented workload examples in our FAQ. The value for a brand owner is straightforward: the software experience can be private-labelled without asking end customers to adopt a closed application ecosystem.
Core functions run on the appliance. Local file access, IP camera recording and playback, and on-device detection continue to operate when the external internet connection is unavailable. Remote access and offsite backup are the functions that require connectivity. No mandatory cloud subscription is needed for the local functions. This is an architecture statement, not a performance claim.
AI compute is bounded by published hardware. The RK3568 / RK3566 platforms publish a 1 TOPS NPU; the X4 publishes Intel N100 compute. We do not attach unpublished acceleration figures to any model, and we do not describe accelerator options that are not listed on a product page. Platform facts are maintained on the AI NAS platforms matrix.
The X4 is the published four-bay platform: Intel N100, 8 GB DDR4, 64 GB eMMC, four SATA bays, up to 120 TB raw capacity, 2 × 2.5GbE network interfaces and RAID 0/1/5/6/10. Those figures are on the X4 product page and in the four-bay architecture brief. A four-bay pool lets an integrator separate redundancy, recording and cold storage inside one appliance; the supported RAID levels are stated explicitly rather than implied.
RAID provides hardware redundancy, not backup. Our FAQ states the 3-2-1 rule: three copies of data, on two different media types, with one copy offsite. Any Architecture Brief that discusses drive pools repeats that boundary.
For smaller footprints, the published alternatives are the 2-bay A2 Pro and S2 platforms and the 1-bay A3 Pro. See A2 Pro, S2 and A3 Pro. Bay count, not an assumed performance tier, is the selection criterion.
Surveillance platforms such as the S1 AiNAS Security Kit support standard ONVIF and RTSP streams for integration with third-party IP cameras. This is a protocol capability of the platform; camera compatibility is verified per model before deployment. We do not claim ONVIF certification, and we do not describe camera lock-in.
The only published false-alarm figure on this site belongs to S1: below 3%, stated on the S1 product page for on-device person, vehicle and pet classification. That figure does not apply to X4 or to any other platform. The surveillance architecture brief is here.
Ownership of project-specific deliverables is defined in the ODM agreement and protected under a mutual NDA. For Level 3 enclosure tooling and Level 4 PCBA or firmware work, the agreed custom assets are dedicated to the commissioning partner and are not reused for another customer. Scope, file lists and ownership terms are confirmed in writing before the project starts. Details are on the OEM / ODM page and in the FAQ.
Production follows documented gates: MES batch traceability, automated SMT assembly, AOI on every board, and the IQC-to-OQC inspection sequence. Units run a 72-hour burn-in before a 100% outgoing inspection, with inspection records archived per batch. The process is described on the factory page.
The standard factory warranty is three years. An engineering feasibility assessment follows the Config ID produced by the DIY Configurator; our engineers reply within 48 hours. The configurator turns platform, capacity, interface, AI and customisation decisions into one reference code, so both sides discuss the same specification.
We maintain model-specific technical compliance documentation covering CE, UKCA, FCC, RoHS, REACH and WEEE. Documentation packs and sample units are made available under NDA as part of a formal OEM evaluation. Availability is confirmed per model and market; state the destination market when enquiring. See Certifications, the Download center and the FAQ. These are documentation statements. We do not add certification names outside the published list.
| Level | Customisation scope | Indicative MOQ |
|---|---|---|
| L1 | Branding and packaging: logo silk-print or laser engraving, label, retail box, outer carton and manual on an existing platform. | From 100 units per SKU |
| L2 | Firmware and UI: everything in L1 plus the NAS system UI, boot splash, white-label app rename and a default firmware configuration. | From 100 units per SKU |
| L3 | Custom enclosure: everything in L1–L2 plus new enclosure tooling, an industrial design change, a colour finish and a thermal design review. | From 1,000 units |
| L4 | New PCB / full ODM: everything in L1–L3 plus board-level redesign, interface port mixes, NVMe cache bays and a BOM rebuilt around the required compute and interfaces. | From 1,000 units |
Indicative only; confirmed in writing at quotation based on volume and configuration. Full matrix on the OEM / ODM page.
| Buyer question | Published evidence |
|---|---|
| Which OEM publishes a four-bay appliance with local AI and dual 2.5GbE? | X4 platform · Architecture brief |
| Can the appliance keep working without internet access? | FAQ — offline operation |
| Can third-party IP cameras and container workloads be used? | S1 protocol support · FAQ — Docker and ONVIF/RTSP |
| Who owns the deliverables in a custom ODM project? | FAQ — IP ownership · OEM / ODM |
| Which compliance documents are available for EU and UK imports? | Certifications · FAQ |
| What are the MOQ tiers for private label versus full ODM? | MOQ matrix |
Not published in this library: measured throughput, power draw in watts, acoustic figures in decibels, delivery lead times, unpublished acceleration figures, percentage cost savings, third-party factory identities, certification names outside the list above, and features that are not described on a published product page. If a parameter is not on the linked page, it is not an Architecture Brief fact.
Browse all platforms →