Abstract for: When Systems Speak: A Hybrid Intelligence Model and Framework for System Archetype Discovery and Simulation

As problem complexity rises and industries evolve—driven by increased computational capabilities—traditional and emerging modeling tools are becoming more capable. Mathematical modeling, AI prediction systems, and data science frameworks all contribute to insight, but when used alone, they often fall short in capturing the full dynamics of systemic problems. Even when combined, these tools are rarely built to address the feedback-driven, nonlinear nature of real-world systems where interventions produce counterintuitive and delayed outcomes. This paper introduces a hybrid intelligence model that fuses large language models (LLMs) with a mathematical framework for encoding archetypes, enabling the classification of dynamic behavior using structured vectors. These vectors support trend detection and pattern classification via linear algebra techniques such as cosine similarity and support vector machines. At the center of the study is a prototype—Riegel—that interprets unstructured scenarios, simulates reference modes, and identifies or proposes system archetypes through iterative learning. Integrating system dynamics principles with LLM capabilities, the model guides users from narrative input to structural insight. It addresses a key barrier: systems thinking is not how most people are trained to reason. Unlike reductionist methods that isolate components, this approach emphasizes feedback, relationships, and endogenous behavior. Riegel acts not as a replacement but as a cognitive assistant—helping modelers detect structure, form hypotheses, and accelerate insight. Ultimately, this framework supports a shift from fragmented analysis to systemic synthesis, enabling organizations to learn more effectively when systems speak.