Architecture, Standards & Jurisdictions
The deeper technical reference for the engine summarized on the homepage — where it sits in an industrial reference architecture, exactly which standard applies to which claim, and jurisdiction-by-jurisdiction coverage. This page carries the depth so the homepage doesn't have to.
Mapped to RAMI 4.0, not bolted onto it
RAMI 4.0 — Germany's Reference Architecture Model for Industrie 4.0 — defines how industrial systems relate across hierarchy, lifecycle, and function. Kaelox is positioned deliberately within it:
P→D Bridge & HITL Gate
Enforcement logic operates here — evaluating and gating AI-influenced decisions before they propagate downward.
ALCOA+ Chain & AEMS
The governance record lives here — structured, queryable, and the source of truth for everything above it.
Configure Mothership
Deployment is scoped to the Station and Work Centre hierarchy levels RAMI 4.0 defines — facility by facility.
Which standard applies to which claim
Kaelox is built toward several distinct standards, each covering a different question. Stated separately, on purpose, so a claim about cybersecurity is never mistaken for a claim about functional safety.
| Standard | Question it answers | Status |
|---|---|---|
| IEC 61508 SIL3 | Functional safety of the software safety core (a COTS component reported as TÜV Süd-certified) | Vendor-disclosed — licensing and scope being independently verified before this claim is repeated externally |
| IEC 62443 SL2 (SDLA + CSA) | Cybersecurity of the platform itself — the deep-tech/AI threat model, SBOM, secure development lifecycle | Targeted, assessment not started |
| ISO 26262 / ISO-IEC TR 5469 | Automotive functional safety and AI-specific guidance, relevant to robotics/edge deployments | Reference framework for Phase I robotics work |
| 21 CFR Part 11 / ALCOA+ | Electronic records and audit-trail integrity | Enforced today, built into the chain |
| FDA QMSR / GAMP 5 | Quality management system and validation evidence structure | Structured output generated today |
| EU AI Act, Article 15 | Adversarial robustness and explainability for high-risk AI systems | Applicability date under active reconciliation — see Standards Registry note below |
Open item: the EU AI Act Article 15 applicability date has conflicting public guidance (Digital Omnibus carve-out status). Logged as unresolved in our internal Standards Registry; not blocking any shipped requirement, and stated here rather than rounded to whichever date is more convenient.
One chain. Every jurisdiction it touches, governed correctly.
The Multi-Jurisdiction Compliance Engine reads a single configuration flag at CONFIGURE and changes what's enforced — applicable standards, egress policy, and audit format — without changing the underlying chain.