The Military demo includes a Phone UI web application that simulates how a field or operations client would communicate with Amnesis under mission constraints. This simulator can be opened from the demo interface and shared as a direct URL, providing controlled access to a live, read-only session. Each session is isolated, time-limited, and restricted to predefined capabilities, ensuring that interaction remains within the boundaries of the demo environment.
In this context, the avatar represents a constrained operational interface whose role is limited to coordination, verification, and compliance enforcement. It does not provide strategic or tactical recommendations and does not attempt to resolve uncertainty through inference. All responses are generated through deterministic queries against the Military workspace and are evaluated against policy, authorization, and temporal context. If a request falls outside these constraints, the system refuses explicitly and provides a reason tied to the applicable boundary, such as missing authorization, insufficient data, or policy restriction.
The simulator includes system-generated operational messages that represent incoming updates, such as changes in constraints, new intelligence summaries, or status notifications. These messages are part of the structured dataset or predefined scenario events and are not produced dynamically by the model. The user may query these updates, request validation, or ask for clarification, and the avatar will respond by referencing the underlying knowledge graph without introducing interpretation beyond what is supported.
The purpose of this interface is to demonstrate how Amnesis behaves when deployed in environments where correctness, predictability, and refusal are more important than flexibility. The interaction model is intentionally strict. Every response is grounded, every limitation is visible, and every action is auditable. The simulator does not attempt to emulate autonomy or decision-making authority. It exists to show how a real system would enforce constraints while still providing access to verified knowledge in time-sensitive situations.