Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2511.13865

Randomized Controlled Trials for Conditional Access Optimization Agent

James Bono·Beibei Cheng·Joaquin Lozano
View original source ↗Read source document ↗Export metadata ↓

Abstract

AI agents are increasingly deployed to automate complex enterprise workflows, yet evidence of their effectiveness in identity governance is limited. We report results from the first randomized controlled trial (RCT) evaluating an AI agent for Conditional Access (CA) policy management in Microsoft Entra. The agent assists with four high-value tasks: policy merging, Zero-Trust baseline gap detection, phased rollout planning, and user-policy alignment. In a production-grade environment, 162 identity administrators were randomly assigned to a control group (no agent) or treatment group (agent-assisted) and asked to perform these tasks. Agent access produced substantial gains: accuracy improved by 48% and task completion time decreased by 43% while holding accuracy constant. The largest benefits emerged on cognitively demanding tasks such as baseline gap detection. These findings demonstrate that purpose-built AI agents can significantly enhance both speed and accuracy in identity administration.

Explore related subjects

econ.GNcs.AIq-fin.EC

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

James Bono, Beibei Cheng, Joaquin Lozano. 2025-11-17. Randomized Controlled Trials for Conditional Access Optimization Agent. https://arxiv.org/abs/2511.13865

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

The time interpretation of expected utility theory

Economic models often maximise expectation values of wealth or utility. In non-ergodic settings, these can differ from time-averages, so that maximising expected outcomes need not maximise – and can systematically reduce – long-run wealth or utility. Ergodicity economics highlights this problem and models individual agents as maximising wealth in the long run, known as growth optimality. Two instances where expected utility maximisation maps to growth optimality are known: linear utility does this for additive wealth dynamics; and logarithmic utility for multiplicative wealth dynamics. Here we show that the mapping holds more generally when the utility function coincides with the ergodicity transformation in the growth optimal model. This mapping offers a theoretical basis for choosing utility functions and suggests the testable hypothesis that wealth dynamics are predictive of risk preferences.

econ.GN↗
arXiv2026

Monetary Regimes and Trade before the Classical Gold Standard: Evidence from the Latin Monetary Union

This paper reexamines the trade effects of the Latin Monetary Union (LMU), a 19th century agreement to standardize gold and silver coinage among several European countries. The LMU provides a useful setting for studying whether monetary arrangements fostered trade before the classical gold standard, when gold, silver, bimetallic, and paper regimes coexisted. Because some countries already shared other monetary standards, treating all non-member pairs as a single control group mixes pairs with and without alternative forms of monetary coordination. I classify pairs by standard and estimate the LMU effect relative to pairs without a common standard, bringing the comparison closer to those used in the literature on the gold standard and contemporary currency unions. The results suggest that the LMU increased trade between its members by approximately 30\% during its early years, when bimetallism was still credible. These effects subsequently faded, converging to zero by the end of the 1870s. More broadly, these findings also highlight the importance of accounting for the existing monetary regimes when estimating the trade effects of other international policies.

econ.GN↗
arXiv2026

Access to Live AI Advice and Behavior Under Risk: An Incentivized Experiment

Generative AI has become an everyday advisor, and the systems people consult are live and interactive, not pre-scripted. We ask whether access to such a system changes behavior under risk. In an incentivized experiment (N = 158), participants made lottery choices with an optional decision aid presented as a conventional pre-written tool, a live one-shot AI, or a live interactive AI they could query, with information format held equivalent across conditions. Risk preferences are elicited via DOSE. We find no evidence that access to a live AI advisor changes risk aversion.

econ.GN↗