Doxa – Economical-constrainted ABM simulator with Ollama (github.com)

Doxa has launched a multi-agent simulation platform designed to model complex economic and social systems using a YAML-driven structure. This innovative platform integrates large language model (LLM) agents, market microstructures, and social dynamics through a FastAPI API and React client, making it accessible for local development and research. Doxa's architecture allows for the simulation of various scenarios, such as resource scarcity and misinformation diffusion, enabling researchers to explore economic experiments and social contagion dynamics in a comprehensive manner. Significant for the AI/ML community, Doxa stands out by providing a unified framework that combines agent-based modeling, generative AI, and economic simulation, unlike existing platforms that often lack cohesive integration. Its modular design, established through a declarative YAML schema, enables users to easily create and validate scenarios while facilitating advanced LLM reasoning and interaction. Key implications include the potential for research applications in policy evaluation and AI-driven negotiations, alongside the promise of future enhancements such as distributed simulation capabilities and improved agent cognition.