Managing PFAS contamination requires water utilities to balance regulatory compliance, treatment costs, and operational constraints. Through a comprehensive treatability study, Louisville Water recently identified powdered activated carbon (PAC) treatment as a best practical strategy for achieving its PFAS compliance goals. To further optimize PAC treatment, this study developed a quantitative risk assessment framework to guide treatment decisions in response to PFOA concentration variation in the source water. Louisville Water has been collecting PFOA data since 2016. Using this decade of data, we employed statistical modeling to estimate the probability of exceeding proposed regulatory thresholds under various treatment scenarios. We employed two statistical approaches. First, we fit lognormal distributions to the data to compute the likelihood of a single PFOA occurrence exceeding the running annual average (RAA) under various scenarios. Second, we conducted Monte Carlo simulations to calculate empirical probabilities of exceeding the RAA based on our own data.   AI-assisted programming accelerated model development, allowing us to rapidly test different scenarios. We generated custom code for both analytical probability calculations and empirical simulations. We validated the results for every scenario by personally checking the generated code, comparing predicted probabilities against observed frequencies, and using a second AI to re-calculate and confirm the original calculations.  A key innovation was modeling treatment effectiveness through scaling factors that simulate PAC removal efficiency. By applying factors of 1.0 (0% removal) down to 0.5 (50% removal) to the data, we quantified how different treatment levels affect compliance risk. Results provided actionable insights for plant operations. For each treatment scenario, we calculated specific exceedance probabilities, enabling managers to select PAC dosing strategies that achieve desired confidence levels for regulatory compliance. Further, our analysis identified treatment intensity thresholds where incremental risk reduction diminishes, which informs cost-benefit decisions about PAC application rates.  This methodology demonstrates how utilities can leverage AI and statistical modeling to make more informed treatment decisions. The approach is transferable to other PFAS compounds, contaminants, and treatment technologies, providing a template for risk-based water quality management. By combining traditional water treatment expertise with modern data science techniques, utilities can optimize treatment processes while managing compliance risk in an increasingly complex regulatory environment.