Recent advances have made generative Artificial Intelligence (AI) systems widely accessible through applications such as ChatGPT, Claude, Google Gemini, and Microsoft Copilot, enabling the generation of text on a wide range of subjects and levels of complexity. At the same time, participatory methodologies used in decision-making processes and futures studies face challenges such as limited access to experts, geographical constraints, and time limitations. In this context, the ability of generative AI to emulate specialized knowledge and produce consistent responses emerges as a promising alternative to support the execution of such methods. Given this scenario, this work aims to investigate and operationalize an approach for integrating generative AI agents into participatory methods. The research was conducted following the Design Science Research methodology, involving the proposal, development, and evaluation of an artifact that enables the configuration and use of these agents in structured interaction processes. As a form of validation, the approach was applied to the Delphi method, operationalized through TIAMAT, a decision support system designed to support Future-oriented Technology Analysis (FTA) methods. The results indicate that the simulated experts are capable of producing coherent assessments and convergence patterns compatible with those observed in panels composed of human experts. These findings highlight the potential of generative AI as structured support for conducting participatory methods in contexts where the mobilization of human experts is limited.
.png)
.png)
