Prototyping an AI-powered Tool for Energy Efficiency in New Zealand Homes

A prototype artificial intelligence tool designed to guide household energy-efficiency decisions in New Zealand has been evaluated by domain experts, returning strong usability scores and signaling a potential bridge between national retrofit programs and individual homeowner action, according to a preprint posted to arXiv [1]. The study, authored by Abdollah Baghaei Daemei and submitted in September 2025 before a June 2026 revision, details a modular dashboard built with Python and Streamlit [1]. The system ingests household data, flags anomalies, establishes a consumption baseline, and simulates scenarios such as LED retrofits or insulation upgrades [1]. Fifteen building scientists, consultants, and policy practitioners assessed the prototype through semi-structured interviews [1]. The tool recorded a mean usability score of 4.3 and a mean value-of-scenario-outputs score of 4.5 [1]. The first submission weighed 460 KB; the second, revised version grew to 961 KB [1]. New Zealand’s housing stock has long struggled with inadequate insulation and inefficient heating, contributing to what researchers describe as widespread energy hardship [1]. Government interventions—the Warmer Kiwi Homes program, the Healthy Homes Standards, and upgrades to the H1 Building Code—have delivered measurable health and comfort gains [1]. Yet the study notes that many retrofits remain partial, household performance data are scarce, and decision-making support for homeowners is fragmented [1]. The AI tool attempts to close that gap by translating national policies into personalized, household-level guidance [1]. The research arrives as open energy-system models gain traction among regulators and system operators in Europe and North America, who began adopting them for planning purposes in the early 2020s [6]. Open models and open data are increasingly used by government agencies to shape net-zero public policy, and engineering consultancies are likewise incorporating them into analysis [6]. The New Zealand prototype extends this trend into the residential sector, offering a replicable framework that the authors argue can reduce energy hardship, improve health outcomes, and support climate goals [1]. The preprint was posted on arXiv, the open-access e-print archive that began as a resource for the physics community and has since expanded across disciplines [10]. EarthArXiv, a similar preprint server launched in October 2017, serves the earth sciences with the same model of rapid, free dissemination before peer review [9]. The New Zealand study’s authors recommend future work on carbon metrics, tariff modeling, integration with national datasets, and longitudinal trials to measure real-world adoption [1].