AI-powered policy frame intelligence for complex regulatory environments
Source document: Arizona Energy Promise Taskforce - Report to Governor Katie Hobbs (March 1, 2026)
VARONA stands for Verifiable Analysis of Regulatory Objects, Narratives, and Actors. It is an AI-enabled policy intelligence platform designed to help researchers, analysts, and decision-makers explore complex regulatory documents.
By combining Large Language Models (LLMs) with ontology-constrained information extraction, VARONA transforms unstructured text into an interactive network of policy frames, actors, regulatory objects, and supporting evidence.
Rather than replacing careful document review, VARONA accelerates it. Every AI-generated classification is linked directly to the original source text, enabling users to verify how the generative AI model interpreted each statement and to assess the evidence independently.
The unit of analysis is the policy frame.
Policy documents contain many different kinds of statements. Some describe regulatory problems, others recommend actions, identify beneficiaries or affected actors, discuss implementation challenges, present supporting evidence, or compare competing priorities.
VARONA refers to these analytical units as policy frames.
Each frame represents a distinct policy-relevant claim extracted from the source document.
The interactive network visualizes relationships among policy frames and the entities they reference:
Nodes represent policy frames and the entities associated with them.
Edges represent analytical relationships extracted from the document.
Colors indicate the type of policy frame.
Filters allow users to restrict the visualization by document section, frame type, relationship type, confidence level, or actor role.
Selecting a node displays the original evidence, model rationale, document location, and the surrounding source context.
Every policy frame shown in VARONA is linked to:
The original evidence excerpt.
Its location within the source document.
The surrounding textual context.
The model-generated rationale.
The confidence assigned during extraction.
This design emphasizes transparency and supports independent verification of every AI-generated classification.
VARONA is intended as an analytical aid for regulatory intelligence, policy research, stakeholder analysis, and document exploration. It should complement—not replace—expert interpretation of policy documents.
Part of The Macumba Manifold — AI for Policy Data Science:
VARONA: Regulatory Objects
Simonizer: Causal Beliefs
Racconto: Policy Controversies
PONder: Purpose-Oriented Networks
Feel free to contact Dr. Edwin Alvarado-Mena through AlvaradoCSS.com.
