EmoMAS: Emotion-Aware Multi-Agent System for High-Stakes Edge-Deployable Negotiation with Bayesian Orchestration

EmoMAS represents a sophisticated research-oriented approach to a very specific problem: making small models (SLMs) act as competent negotiators by using Bayesian methods to model emotional dynamics. While the technical premise—combining Bayesian orchestration with emotional intelligence—is a 'novel combination' that solves a real gap in SLM capabilities, the project currently lacks any significant moat. With 0 stars and only a few forks, it is effectively a fresh academic reference implementation. Its defensibility is low because the competitive advantage lies in the mathematical approach (Bayesian modeling) rather than a network effect, proprietary dataset, or robust software ecosystem. Frontier risk is low because major labs (OpenAI/Anthropic) are focused on making larger models more efficient or developing generalized agentic reasoning; they are unlikely to build niche emotion-aware negotiation frameworks specifically for edge/rescue robots. However, platform risk is medium because mobile OS providers (Apple/Google) could integrate similar 'emotional reasoning' layers into their on-device SLM stacks (e.g., Apple Intelligence). The displacement horizon is 1-2 years, as the inherent reasoning capabilities of 8B-class models are advancing so rapidly that the need for complex external Bayesian orchestration to handle emotions may be 'eaten' by better model weights in the near future. Competitors include general multi-agent frameworks like AutoGen or CrewAI, though they currently lack the specialized Bayesian emotional logic found here.