The internet leaks unverified information into every answer you read.
CCI is your cognitive companion — online, with a language model, or without.
The Convergent Cognitive Intelligence framework is a code-deployed mathematical reasoning architecture authored by Marisol Guzman Montoya, an independent researcher in geophysics, wave theory, and cognitive architecture based in Oslo.
The framework was first authored on 13 May 2024 and is built from twenty-five years of independent research and the inventor's own authored equations.
CCI is not a prompt. It is not a chatbot. It is the first cognitive architecture whose validation layer is mathematics, not statistics.
A public position paper establishing the framework, its intellectual lineage, and its research program.
We introduce Convergent Cognitive Intelligence (CCI), a mathematical reasoning architecture designed to serve as a cognitive companion for any human professional in any domain. Unlike statistical language models, which approximate truth through pattern correlation across training corpora, CCI applies a mathematically-grounded validation layer that decomposes a query into the domains it implicates, validates each claim against named authoritative sources, and produces a structured audit trail of the reasoning process.
The architecture treats large language model outputs as hypotheses rather than authoritative answers, and iterates a convergence procedure until either internal coherence is achieved or the system reports an honest non-convergence with named gaps. The mathematical foundation of CCI is derived from the inventor's prior independent work on convergence dynamics in non-linear systems, reinterpreted as abstract operators on structured reasoning representations.
This paper establishes the conceptual framework, the intellectual lineage, the public capability descriptors, and the research program of CCI. Architectural mechanisms and mathematical specifications remain under confidentiality and are available for licensing under appropriate commercial agreements.
Authorship of the CCI architecture commenced 13 May 2024. The mathematical foundation derives from an earlier unpublished manuscript by the inventor on convergence dynamics in non-linear systems, held in escrow with the inventor's legal counsel. The full position paper is now publicly available on Zenodo under CC BY 4.0: From intention to semantic harvesting through structural deterministic reasoning (September 2026).
Convergent Cognitive Intelligence is a deterministic reasoning architecture external to any LLM. It operates on top of the language model, sends a structure to the LLM, and the LLM contributes semantic content inside that structure. CCI's governance operator, domain decomposition, claim tiering, and audit trail run the same deterministic logic regardless of which LLM is called. The October 10 release documents three findings on a biology diagnostic subset.
This benchmark does not claim CCI produces more correct answers than frontier LLMs, nor that the findings generalize beyond the biology diagnostic subset tested here. Cross-domain generalization is deferred to the December 1, 2026 NL4OPT release. The LLM-as-judge content correctness score is judge-estimated with documented biases, disclosed in the methodology addendum. The methodology is open; the CCI implementation and the convergence operator mathematics remain the author's proprietary work.
Twenty-five years of independent research across geoscience, wave theory, and non-linear systems. CCI is the result of a separate body of mathematical work, developed independently, that translates the inventor's own authored equations into a cognitive validation architecture.