Human Agent Interaction
Defining how humans interact with agents
Purpose and Structure of this Guide
Comprehensive, practical, and expert-level resource based on the five core HAX principles.
Dedicated Boundaries: Ensuring AI Operates Within Defined Limits
Each principle guides AI agents to operate within defined boundaries set by users. Using scoped access, constrained prompts, and permissions, it prevents overreach.
Evaluation includes tracking violations, user feedback, and metrics to ensure safe, aligned behavior.
Guiding Principles
This guide focuses on five core HAX principles derived from these guidelines:
