New framework proposes publisher-side knowledge engineering model for AI search, answer engines and agentic systems
PRESCOTT, Ariz., Aug. 30, 2026 — The unabridged edition of The Visibility Code™, a public framework for Knowledge Engineering for Answer Engines, has been released at VisibilityCode.com.
The framework addresses a publishing problem created by the growing role of artificial intelligence systems in retrieving, interpreting and synthesizing information before users reach an originating web page.
Traditional digital publishing developed around a model in which publishers created documents, search engines discovered and ranked them, and people performed much of the interpretation. Answer engines, language models and agentic systems increasingly perform part of that interpretive work themselves, including identifying entities, retrieving facts, comparing information and assembling answers.
The Visibility Code argues that this shift changes the publisher’s job.
“Publishers routinely know more about their information than they explicitly publish,” said David W. Bynon, author of The Visibility Code and inventor of the WebMEM® Protocol. “They know which entity a fact describes, where it came from, when it applies, how it relates to other information and which conditions limit its meaning. Much of that structure disappears when the information is reduced to a conventional web page.”
The framework proposes that publishers treat this missing structure as a knowledge representation problem rather than solely as a search-ranking or content-optimization problem.
Its central principle is straightforward: Publishers should explicitly publish the semantic structure they already know.
From document retrieval to knowledge resolution
The framework distinguishes retrieval — finding potentially relevant information — from resolution, the process of determining what information means and whether it applies to a particular entity, context or information need.
The current Version 2.0 framework covers Semantic Data Templates, machine-facing knowledge fragments, facts and assertions, identity, provenance, relationships, applicability, resolution, resolver architecture, serialization and conformance.
It also establishes a boundary between publisher-controlled information and the behavior of proprietary AI systems.
Publishers can control what knowledge they expose and how explicitly they represent its identity, provenance, relationships and applicability. They cannot control whether a search engine, answer engine, language model or agent retrieves, ranks, cites, trusts or uses that information.
That distinction is intended to separate observable publishing practices from unsupported claims about the internal operation of proprietary AI systems.
Knowledge Engineering for Answer Engines
The framework describes the emerging discipline as Knowledge Engineering for Answer Engines: applying knowledge representation, information architecture, provenance, resolution and governance principles to public web publishing.
The approach extends beyond conventional measures of AI visibility such as brand mentions, citations and share of voice.
Under the framework, a source may be visible but still be represented incorrectly. A citation may point to an authoritative source while an important qualification is omitted. Current and obsolete information may coexist. An entity may be mentioned but incorrectly resolved.
The framework therefore treats retrieval, resolution, representation fidelity, attribution, temporal validity and task utility as distinct dimensions that publishers can observe and evaluate.
Two-Tier Publishing and WebMEM
A central publishing model within The Visibility Code is Two-Tier Publishing, in which conventional human-facing content is accompanied by a complementary machine-facing representation of publisher-known knowledge.
WebMEM, a publisher-side knowledge representation protocol developed from the same research, provides one implementation architecture for that machine-facing layer.
The current WebMEM model uses page-level Semantic Data Templates and modular knowledge fragments to explicitly represent information such as identity, facts, provenance, relationships, definitions, indexes and derived information. The framework expressly limits WebMEM to the publisher side of the system boundary and does not prescribe how consuming AI systems retrieve, rank, reason over or synthesize that information.
Developed through production publishing research
The framework evolved from work originally published in July 2025 and was substantially revised during 2026 following production implementation. According to the published revision history, that work led to a shift away from earlier concepts centered on influencing machine memory or retrieval behavior and toward publisher-controlled knowledge representation.
Production research supporting the framework includes work on MedicarePlans.com, where machine-facing knowledge structures are being deployed and observed in a live public-web environment.
The research methodology separates publishing interventions from subsequent observations so that changes in search or answer-engine behavior are not presented as evidence of undocumented internal mechanisms.
Public unabridged edition
The Visibility Code — Unabridged is being maintained as a public reference work rather than a conventional blog or series of independent articles.
The unabridged framework includes 19 chapters plus a prologue and epilogue covering machine knowledge, WebMEM, semantic data templates, retrieval and resolution, provenance and governance, measurement, publisher feedback loops, representation optimization, temporal knowledge integrity, misinformation resilience, interoperability and the future of AI visibility.
The public edition is available at:
The Visibility Code — Unabridged
About The Visibility Code
The Visibility Code is a framework for Knowledge Engineering for Answer Engines developed by David W. Bynon. It examines how publishers can make public knowledge more explicit, resolvable, attributable, applicable and maintainable as search engines, answer engines, language models and agents increasingly interpret and synthesize information.
The framework’s guiding boundary is: Optimize what you publish. Measure what the machine reflects.