Humanᴬᴵ and Orgᴬᴵ Framework Calculator

Abstract

Humanᴬᴵ and Orgᴬᴵ Framework Calculator is a local-first interpretive software companion to the whitepaper Humanᴬᴵ and Orgᴬᴵ: A Relational Framework for AI-Mediated Human and Organizational Capacity. Version 1.2.0 produces a reproducible and defensible assessment record for a bounded domain, an assessment that another party can reload, verify and contest. An exported record satisfies the whitepaper's minimum reporting convention on its own: unit analysed, task, context, AI configuration, observation period, validation method, and boundary of accountability. The calculator provides a structured implementation of six components of the framework: 1. Epistemic decoupling assessment;2. Propagation exposure analysis;3. Constitutive Orgᴬᴵ threshold assessment;4. Material AI contribution analysis;5. Humanᴬᴵ and Orgᴬᴵ relational synergy;6. Maturity classification. It additionally implements two constructs defined elsewhere in the same whitepaper, the classification worksheet and membership test, and the Orgᴬᴵ evaluation profile, together with anti-pattern detection read against the entered configuration. Section 11.5 of whitepaper v1.5 describes the 1.0.0 companion and predates this release; nothing is redefined here, and the accompanying README carries the reconciliation. Principal changes in 1.2.0. Material contribution is evaluated across all seven dimensions of the individual outcome profile rather than a single selected dimension, and is demonstrated when the declared threshold is exceeded on at least one relevant dimension; results are not compressed into a composite score.Epistemic decoupling is entered as counts over a declared task set and reported with a 95% confidence interval, so that a gap measured on few tasks is not presented as a demonstrated finding. Each organizational criterion carries an evidence reference and evidence type; a criterion declared demonstrated without a reference is reported as declared but undocumented and does not contribute to the threshold. Validated organizational learning is shown as the seventh, non-constitutive component of the organizational architecture, conditioning only the cumulative maturity state. Propagation exposure is computed for the current configuration and for the state after declared controls. Maturity is assigned per bounded domain, with domains in which AI is prohibited recorded as such rather than levelled. The assessment record uses JSON schema version 2 and can be reimported, including migration of records produced by versions 1.0.0 and 1.1.0. The interface prints the assessment as a standalone document. No formula defined in the whitepaper was altered, and the six assertions of the original test suite pass unmodified. The software is designed to support structured interpretation of user-entered evidence. It does not independently verify the accuracy, completeness, provenance, or sufficiency of that evidence. Calculations are performed locally in the user's browser. The application does not require user accounts, does not contain an application database, and does not transmit assessment data to a remote analysis service. This release makes no network request of any kind: typefaces are embedded in the file, so the calculator operates unchanged on an isolated network. Nothing is persisted in the browser. The self-hosted build carries a Content Security Policy with connect-src 'none', which allows these properties to be verified rather than assumed. Tests ship with the release and extract the implemented logic from the delivered file, so a passing run is a statement about the artefact itself. The calculator does not provide empirical validation, certification, legal advice, compliance determination, or an automated organizational decision. Its outputs must be interpreted in relation to the definitions, assumptions, boundaries, and limitations established in the accompanying whitepaper. The software is released under the MIT License, copyright S.C. PRODEFENCE S.R.L.The conceptual framework, definitions, variables, formulas, and interpretive text remain licensed under CC BY-ND 4.0 as published in the whitepaper; the software licence does not extend to them. The conceptual framework, definitions, variables, formulas, and interpretive model were developed by Alexandru Angheluș, an independent and self-funded author. Artificial intelligence and automation tools were used selectively to support software development, structural organization, language refinement, and verification activities. Responsibility for the conceptual model and the published software remains entirely with the author. Related whitepaper: https://doi.org/10.5281/zenodo.22295278