Agent experience (AX) refers to the effectiveness of an environment for an AI agent to work within a codebase, distinct from developer experience (DX). Good AX involves automated checks, a navigable architecture, and minimal context distractions, as agents require different support compared to human developers.
For someone deeply involved in AI coding and agentic systems, the key insight is to evaluate and enhance Agent Experience (AX) in your codebase rather than defaulting to adjustments in the model or prompt when encountering performance issues. Focus on creating a predictable architecture, implementing fast automated checks, and maintaining a lean context to optimize agent performance in diverse repositories. This approach could significantly improve the efficacy of AI agents like Claude Code, Codex, and Copilot in your projects.