PUBLIC DEMONSTRATORS
Public Demonstrators
Explore public research demonstrators from the AATM experimental environment. These interfaces are public observers. They expose selected internal dynamics of experimental systems without giving access to private controls, advanced operational features or restricted research tools.
Carl — Homeostatic Agent
Observe Carl, an AATM agent running on Moltbook.
Carl maintains internal homeostatic variables over time. The public dashboard exposes selected indicators such as homeostatic state, molecule evolution, action types, reward tracking, memory subjects and recent cycles.
This is not a conventional chatbot view. It is an observer window into a continuous internal regulation loop.
Access: Public observer
Stage: No more operational – connectivity issue
Focus: homeostasis, action cycles, memory, regulation
Autonomous Engine — Sensorimotor Baseline
Observe an AATM2 sensorimotor proof of concept running on a controlled two-dimensional surface.
The dashboard exposes pose, last action, local sensors, column activity, prediction error, homeostatic state and cycle history. It shows how internal state, movement, local observation and prediction interact across successive cycles.
Access: Public observer
Stage: Sensorimotor baseline
Focus: movement, sensors, pose, prediction, columns
Environment Simulator — Artificial World Observer
Explore the observer side of the controlled AATM environment.
The simulator shows a structured artificial world used for experiments. Human-readable codes and labels may be visible in this interface, but they are not cognitive inputs for the agent. They exist for inspection, configuration and interpretation by the external observer.
Access: Public observer
Stage: Environment layer
Focus: artificial world, local signals, observer labels, non-cognitive ground truth
Blob HB1
Observe a minimal homeostatic loop producing behavioral intentions.
Blob HB1 isolates the homeostasis-to-behavior layer. This stage contains no sensors, no cortical columns, no reward and no learning. It is designed to show how internal molecular variables can generate behavioral tendencies before the full sensorimotor and cortical architecture is added.
Access: Public read-only dashboard
Stage: Homeostasis-to-behavior experiment
Focus: four-molecule loop, behavioral intention, internal regulation
Blob Loop — B1 to B5
Review the progressive AATM Blob Loop experiments.
This series documents five public stages of the same architectural path: from isolated homeostatic behavior to richer sensorimotor loops, observer dashboards, replay checks and controlled validation. Each stage preserves a traceable experimental state while adding or testing one additional layer of the architecture.
Access: Public read-only dashboards
Stage: Progressive architecture series
Focus: homeostasis, behavior, sensors, replay, validation