About
An open framework for Agent Experience
An open reference for Agent Experience
axd.ax documents Agent Experience, or AX: the discipline of designing websites and digital products so AI agents can find, understand, navigate, and operate them. It is a practical reference for product teams, designers, developers, marketers, and technical leaders adapting the web for a new class of user.
Human visitors interpret visual hierarchy, explore menus, and recover from ambiguity through judgment. Agent behavior varies by model and configuration. Explicit structure, machine-readable meaning, clear actions, predictable states, and useful error paths provide inspectable site evidence, but do not guarantee model behavior or task success. AX asks whether a digital product communicates and behaves well enough for an agent to act with confidence, while preserving the clarity and safeguards people expect.
A framework for evaluating real products
The site organizes AX into principles, patterns, and five measurable dimensions. Discoverability asks whether agents can find and recommend a product in relevant research. Navigability examines whether they can accurately describe what it does, who it serves, and how to begin. Operability covers whether an agent can complete useful tasks through an interface or API.
Recoverability measures whether agents can recognize and correct wrong information or failed actions. Transparency examines whether they can represent a product's scope, limitations, and appropriate use cases accurately. Together, these dimensions form the AX Score, a 0 to 100 assessment with each dimension scored from 0 to 20. The score is a diagnostic for agent readiness, not a claim about business or financial performance.
How to use axd.ax
Start with the principles to understand the design rules behind strong agent experiences. Use the dimensions to identify where a product is clear or fragile. Consult the patterns for implementation approaches, then use the checklist or AX Score to turn the framework into a repeatable review of a live website.
The goal is not to optimize for one crawler or one model. It is to make product information and interactions more explicit across systems: clear semantics, stable navigation, well-defined actions, honest constraints, and recovery paths that help an agent know what happened and what to do next. Those qualities also tend to make digital products easier to maintain and explain to people.
Maintained in the open
AX is a young discipline. The reference documents proposed practices that can be tested against captured site evidence. It is designed to evolve as agent capabilities and web standards change. The vocabulary and scoring model give practitioners a shared starting point for testing, discussing, and improving agent-facing experiences.
axd.ax is maintained by the team at Optagen.ai. Optagen applies AX and AI visibility concepts alongside its broader ecommerce intelligence work. This site remains focused on an open framework and reference. It is not an independent certification body and does not claim recognition by AI providers.
If you want to know how AI agents experience your product today, start with the AX Score. If you want a human-reviewed AXD journey audit, request a scope. Ecommerce operators can also review the optional Optagen handoff.
Related Resources