Receptiviti

Receptiviti develops scientifically grounded measurement infrastructure for human-state-aware AI systems.

Our work focuses on making cognitive and psychological interaction dynamics measurable, observable, and usable within AI evaluation, monitoring, and conversational systems.

Current AI systems adapt to users implicitly through language, but the underlying interaction-state signals influencing model behavior remain largely opaque, inconsistent, and unavailable to the systems meant to evaluate or govern them.

We are interested in approaches that make interaction state explicit, measurable, longitudinally trackable, and inspectable.

Research Areas

• Human-state-aware AI evaluation
• Longitudinal conversational dynamics
• Cognitive load measurement
• Psychological signal extraction from language
• Interaction-state instrumentation
• AI safety and governance
• Conversational system observability
• State-conditioned evaluation frameworks
• Human-AI interaction measurement
• Longitudinal risk and escalation dynamics

Focus

Our current work explores how explicit interaction-state measurement can support:

• Better AI evaluation frameworks
• Safer conversational systems
• Longitudinal interaction monitoring
• Context-aware AI behavior
• Human-centered AI governance
• State-conditioned response adaptation

Notes

Our measurement approaches draw on validated psycholinguistic and behavioral research methods designed for structured observation of interaction dynamics within AI systems.

We are particularly interested in evaluation and observability approaches that treat interaction state as measurable infrastructure rather than latent implicit inference inside model behavior.