What the RNX Protocol Measures
AI visibility problems are often described with soft language: the brand is not showing up, the model does not understand us, the answer is wrong, competitors appear first, the site is not AI-ready. Those statements are useful symptoms, but they are not specific enough to repair. RNX exists to make the problem measurable.
At a high level, the framework separates AI presence into several operating surfaces. Each surface asks a different question. Can the domain be retrieved and understood? Does the wider internet corroborate the entity? Can an agent discover and interact with the business safely? Are the governance and resilience signals strong enough for high-stakes environments?
Domain perception
The 100-series is about what AI systems can see when they inspect the business domain. It includes technical discoverability, canonical consolidation, structured identity, content topology, freshness labeling, headless rendering behavior, geospatial context, and hallucination guardrails.
This layer matters because the website is often the most controllable source of truth. If the site blocks crawlers, hides essential copy behind fragile rendering, splits authority across duplicate URLs, or fails to declare a clear organization identity, AI systems have to do more interpretive work. More interpretive work means more room for drift.
Domain perception also connects to performance and architecture. Fast response paths, useful payloads, and clear internal structure help machine systems extract meaning without spending limited context on boilerplate or navigation noise.
Authority and consensus
The 200-series looks outside the website. It asks whether public sources agree on who the business is, where it operates, what it offers, and why it should be trusted. This includes name, address, phone, and domain consistency, authority node quality, credential provenance, sentiment risk, and contamination from outdated or harmful references.
Authority is not just volume. Ten weak mentions that repeat the same stale fact may be less useful than two independent sources that strongly confirm the right entity. AI systems are sensitive to consensus, but consensus can be polluted. The goal is not to manufacture trust. The goal is to reduce ambiguity and make real trust easier to verify.
Agent readiness
The 300-series is about whether the business can communicate with AI systems in a controlled, machine-readable way. This includes discovery, manifests, deterministic tool contracts, action boundaries, live data, verified state assertions, and audit trails for agentic workflows.
This layer is important because the web is moving from passive pages to active AI-to-AI interaction. A human-facing page can answer some questions, but it does not automatically tell an agent what it may do, what it may not do, what data is current, or what response shape is safe to trust.
Governance, resilience, and future compatibility
The 400-series and 700-series cover the controls that become more important as AI interaction becomes more consequential. These include liability boundaries, attribution, provenance, data residency, crisis protocols, adversarial defense, and compact trust packaging for constrained models.
Not every business needs the same depth on day one. A local service company and a multinational bank should not have identical requirements. But even smaller businesses benefit from understanding where the maturity curve is headed.
What RNX does not do publicly
RNX is a diagnostic language, not a public dump of every internal weighting, threshold, or remediation sequence. The public concepts help teams understand the shape of the problem. The operational details belong inside controlled scans, client dashboards, and implementation work.
That distinction protects both sides. Businesses get a clearer view of why AI systems may skip or misunderstand them. Runexus keeps the deeper machinery where it belongs: inside the product.