Where Survival Is a Conversation
Cognitive Crisis Simulator is a voice-driven, multi-agent simulation platform designed to explore how intelligence behaves under pressure. It creates high-stakes environments—such as failing spacecraft or critical system breakdowns—where AI agents and human participants must communicate, reason, and act in real time to survive. At its core, the platform models decision-making as a social process, where outcomes depend not just on logic, but on communication, trust, and timing.
A defining feature of the system is its dual-layer cognition model, where agents maintain private internal reasoning while expressing selective, strategic speech externally. This enables the emergence of complex behaviors such as deception, persuasion, alliance formation, and conflict. Combined with a theory-of-mind engine, agents attempt to predict how others think and react, allowing for more realistic social dynamics and adaptive strategies during crises.
The platform also incorporates a cinematic simulation layer, including a Director AI that dynamically controls pacing, tension, and narrative flow. Environmental stressors—such as resource scarcity, system failures, and time pressure—are continuously adjusted to test the limits of agent performance. Voice interaction is central, making communication immediate and immersive, and turning every decision into a spoken, consequential act.
Beyond simulation, Cognitive Crisis Simulator is built as a research framework. It supports controlled experiments, batch simulations, and detailed analytics to measure outcomes like survival rates, trust accuracy, deception success, and decision quality. With scenario customization, persistent agent learning, and full data logging, it enables rigorous study of emergent intelligence, cooperative behavior, and failure modes in complex systems.

- Cognitive Crisis Simulator – A voice-driven, multi-agent platform that models how intelligence, communication, and decision-making evolve under high-pressure crisis conditions.
