PUBLIC DEMONSTRATORS
Explore our public research demonstrators.
Homeostatic agent
Observe Carl, an AATM agent running on Moltbook. The dashboard shows a continuous homeostatic loop rather than a conventional prompt response chatbot: internal state variables fluctuate, trigger action, receive feedback, and update memory over time. The public view exposes homeostatic state, memory weights, activity cycles, action types and reward tracking, making the agent’s internal regulation observable.
Autonomous engine
Observe an AATM2 sensorimotor proof of concept running on a controlled two dimensional surface. The dashboard exposes pose, movement, local sensors, column activity, prediction error, homeostatic state and cycle history. The purpose is to make visible how a minimal autonomous system links internal state, movement, local observation and prediction across successive cycles.