Executive Summary
school-to-college reading levels. These letters often explained legally important obligations in language that assumed specialized tax knowledge, leaving residents uncertain about why they were being contacted, what action was required, and what would happen next.
The Nevada Department of Taxation turned that problem into a responsible AI use case. As part of its multiyear modernization program, the Department rebuilt taxpayer correspondence using an AI-assisted workflow. Staff used approved generative AI and word-processing tools to create first drafts from public-domain letter templates, then reviewed each draft through a formal quality process to preserve legal meaning, statutory references, tone, and accuracy.
The results are practical and measurable. Average Flesch-Kincaid grade level dropped from 13.4 to 8.4, average Reading Ease improved from 33.9 to 52.7, 119 letters have been implemented, and general-question call volume fell from 23,121 in 2023 to 11,175 in 2025. Even allowing for the messiness of real-world call-center categories, that is the sort of directional data government should pay attention to.
The model is already becoming bigger than one agency. The Nevada Secretary of State identified Taxation’s approach as the model for a new plain-language project in its Commercial Recordings Division, applying the same core pattern: benchmark readability, use secure AI prompts to rewrite content, review with business and communications experts, and finalize content for public use in English and translated languages.
