Empowers AI agents with real-time human physiological and emotional state awareness to enable adaptive interactions.
Sponsored
Nefesh provides a Model Context Protocol (MCP) and Agent-to-Agent (A2A) server that delivers real-time awareness of human physiological and emotional states to AI agents. By processing sensor data like heart rate, voice, facial expression, and text sentiment, it generates a unified human state and machine-readable suggested actions. Crucially, it offers a unique closed-loop feedback system, `adaptation_effectiveness`, which informs agents whether their previous responses successfully reduced stress, enabling continuous self-improvement without prompt engineering.
Key Features
01Provides real-time human physiological and emotional state awareness from sensor data.
02Offers closed-loop feedback via `adaptation_effectiveness` to measure previous action impact.
03Generates machine-readable `suggested_action` for AI agents to follow directly.
04Supports autonomous API key provisioning by AI agents through an automated process.
05Integrates seamlessly with both Model Context Protocol (MCP) and Agent-to-Agent (A2A) protocols.
060 GitHub stars
Use Cases
01Enabling AI agents to adapt their communication and responses based on a user's real-time emotional and stress levels.
02Building self-improving AI systems that learn optimal interaction strategies through continuous feedback on human state.
03Developing AI applications that proactively adjust their behavior to maintain user engagement and reduce stress.
