LLMs with a Heart: Using Contextual Heartbeat Haptics to Add Feeling to Text and Audio LLM Interactions

Conversational LLMs are physiologically inert, lacking the embodied signals that shape how emotion, urgency, and presence are perceived in human interaction. We introduce a contextually generated haptic heartbeat that is autonomously produced by an LLM based on the affective trajectory of its output and delivered through commodity devices. This work presents an in-progress exploration of heartbeat haptics as a novel multimodal channel for conversational AI, reporting preliminary results from a study evaluating text and audio with heartbeat haptics that provides early directional evidence that LLM-driven heartbeat signals may modulate perceived arousal and valence, enhance emotional expressiveness, and increase credibility and immersion relative to unimodal interactions. These findings are preliminary and should be interpreted conservatively, but they point to a potentially meaningful breakthrough for multimodal interface design: simulated physiological cues may add a new layer of affective embodiment to disembodied systems. We introduce a tightly coupled language-to-haptics pipeline and outline the next steps needed to establish robustness, scalability, and cross-modal coherence, suggesting that LLM-generated heartbeat haptics could open a promising design space for more expressive, credible, and experientially rich conversational agents.