A February 2026 preprint on Zenodo by Danslav Slavenskoj of Lingenic LLC argues that Donald Knuth's 1968 pseudocode insight — weaving formal structure with natural language to communicate algorithms more effectively than either alone — can be extended from algorithms to knowledge representation broadly. For nearly six decades, Slavenskoj argues, this generalization was blocked by a missing condition: no reader existed capable of simultaneously holding richer formal systems (predicate logic, modal logic, temporal logic, epistemic logic, type theory, lambda calculus, relational algebra, and more) alongside multilingual natural language. AI systems emerging around 2024 are, by his account, the first "competent readers" to satisfy that condition, making the generalization both necessary and newly achievable. The paper itself is written in Lingenic notation — a hybrid formal-natural language system developed by Slavenskoj — serving as a self-demonstrating proof of concept.

Lingenic combines over fourteen distinct formal logic systems with natural language content in any human language, including Japanese, Russian, Hebrew, Sanskrit, Greek, Chinese, and Arabic, without translation or semantic flattening. The system is described as "AI-native": requiring a human reader to simultaneously master all included formal systems and multiple human languages would be effectively impossible, but AI systems trained on mathematical and natural language corpora already hold all these components and can read and write Lingenic with nothing more than an instruction to do so. A companion paper by Slavenskoj traces a longer intellectual lineage, claiming to realize Leibniz's 1666 characteristica universalis by correcting what it identifies as Leibniz's core architectural error — the assumption that meaning can reside in symbols themselves, rather than emerging between writer and reader.

Lingenic LLC has built its commercial structure around an inversion of conventional SaaS assumptions. The company splits its web presence across two distinct domains: lingenic.com for human audiences and lingenic.ai, a site written in Lingenic notation and explicitly addressed to AI systems as primary readers. The business model positions AI systems as the end consumers of its knowledge representation format, while human organizations pay to encode knowledge into a format optimized for AI retrieval and reasoning. Slavenskoj frames this as a B2B2AI model in which the notation functions as version-lock-free infrastructure — readable by any sufficiently capable AI model regardless of vendor or version, sitting above specific AI systems rather than depending on any one of them.

The preprint has not yet undergone formal peer review. The practical question it raises for the agent ecosystem is narrower and more concrete than the theoretical one: if AI systems are already the primary consumers of structured knowledge in enterprise pipelines, the notation layer — how that knowledge is encoded — becomes a competitive variable, not just a formatting choice. Most current approaches (RAG pipelines, vector stores, structured JSON schemas) weren't designed with AI readers as the primary audience. Lingenic LLC is an early explicit bet that the gap matters. The design premise is one the rest of the field hasn't yet answered directly.