Simulation Agent: A Framework for Integrating Simulation and Large Language Models for Enhanced Decision-Making

TL;DR

The paper tackles the problem of inaccessible, complex simulations and ungrounded LLMs by introducing a simulation agent framework that merges a robust simulation engine with a LangChain-based AI Agent. The approach enables natural-language interaction to run simulations, modify inputs, and interpret outputs through post-processed JSON summaries, all while grounding insights in verifiable system dynamics. This yields accessible, interpretable, and empirically grounded decision support across domains, reducing hallucinations and improving reliability. The work lays a foundation for broader domain generalization and outlines future directions in enhanced output interpretation, multi-agent architectures, and rigorous framework validation.

Abstract

Simulations, although powerful in accurately replicating real-world systems, often remain inaccessible to non-technical users due to their complexity. Conversely, large language models (LLMs) provide intuitive, language-based interactions but can lack the structured, causal understanding required to reliably model complex real-world dynamics. We introduce our simulation agent framework, a novel approach that integrates the strengths of both simulation models and LLMs. This framework helps empower users by leveraging the conversational capabilities of LLMs to interact seamlessly with sophisticated simulation systems, while simultaneously utilizing the simulations to ground the LLMs in accurate and structured representations of real-world phenomena. This integrated approach helps provide a robust and generalizable foundation for empirical validation and offers broad applicability across diverse domains.

Figures (1)

• Figure 1: A diagram of the framework, illustrating the interactions between its core components.