California’s Community Choice Aggregators (CCAs) stand at a pivotal moment. Serving more than fourteen million residents and managing more than sixty terawatt‑hours of annual demand, they have matured from local initiatives into market‑shaping institutions. Simultaneously, cloud‑based artificial‑intelligence (AI) platforms have become both affordable and exceptionally accurate, offering utilities worldwide new tools to strengthen grid reliability, deepen customer engagement, and reduce operating costs. This white paper explains why the convergence of these developments makes 2025 an optimal time for CCAs to embrace AI. It outlines the most relevant use cases, details pragmatic implementation steps, and highlights the governance measures needed to align technology deployment with public‑service values.

The evaluation of behavioral programs requires rigorous measurement and verification. While the randomized controlled trial (RCT) lies at the core of modern program evaluation, in most situations it is not feasible to implement a randomized approach. This is particularly true when programs have gone full-scale and the success of the program needs to be evaluated outside of the confines of an experimental framework involving the random allocation of households to treatment and control groups. This report aims to describe methods for performing a scientifically sound program evaluation in situations where a properly randomized experiment is not possible.