Voice that belongs to a living system.

Traditional voice AI maps text to sound one response at a time. That is useful for narration, but a social robot needs a continuous operating state: who is present, what just happened, which relationship is active, where attention should move and what action is safe next.

IronHeart.AI treats voice as part of the cognitive loop. Perception, dialogue state, memory, personality, gaze and actions update one another during the interaction. The voice can therefore preserve identity and conversational direction instead of restarting with every generated message.

Robot OEMs and system integrators use this layer to create a governed interaction model for a specific body, environment and role. The result is not a generic chatbot mounted inside hardware. It is runtime infrastructure for embodied social behavior.

A useful embodied system cannot treat voice, memory and action as unrelated calls. IronHeart coordinates the modules around the live person, environment, role and task.

Each module can be configured per robot or product while published versions preserve a reproducible operational state.

Primary buyer: Robot OEMs, social robotics teams, device manufacturers and system integrators building customer-facing embodied AI.

What to measure: Evaluate interruption recovery, turn-taking, identity continuity, attention shifts, response timing and safe action completion—not only transcript quality.

Required controls: Per-robot personality, memory policy, knowledge boundaries, voice identity, gaze behavior and action permissions.