In modern organizations, decision-making rarely relies on a single indicator, as business problems are typically shaped by conflicting requirements such as lower cost, higher quality, lower risk, faster delivery, and greater reliability. In this context, decision support systems (DSS) and business intelligence (BI) provide a benefit framework for structuring problems, processing data, and evaluating alternatives. At the same time, generative AI, especially large language models, is becoming increasingly available as an auxiliary tool for formulating formulas, explaining results, preparing reports, and interpreting findings. However, its use in business decision-making requires transparency, source verification, and clear human accountability.

This paper aims to demonstrate, both theoretically and practically, how multi-criteria decision-making can be implemented as a lightweight DSS model in Microsoft Excel and how generative AI can be incorporated as a supportive, rather than autonomous, component of decision-making. The Simple Additive Weighting / Weighted Sum Model (SAW/WSM) is applied through the normalization of cost and benefit criteria, weighting, and sensitivity analysis based on changes in the weight assigned to price. The case study includes three suppliers and four criteria: price, quality, technical support, and delivery time.

The results show that under the initial criterion weights (0.4, 0.3, 0.2, 0.1), TechSol achieves the highest overall score (0.9027). However, after increasing the weight of price to 0.6 and proportionally rescaling the other weights, GlobalIT becomes the top-ranked supplier (0.9022). This finding confirms that DSS should not be seen as tools that provide a single definitive answer, but rather as transparent frameworks for examining the consequences of different managerial priorities. The paper concludes that an Excel-based MCDM model is a practical, accessible, and reproducible solution for operational decision-making in small and medium-sized organizations, but that its reliability depends on data quality, spreadsheet error control, and the responsible use of generative AI.