Summary of Netlify’s “Agent Experience (AX)” Concept
What is AX?
Agent Experience (AX) refers to the holistic experience that AI agents (software agents) have when interacting with a product, platform, or system — analogous to how User Experience (UX) focuses on humans.
As more AI agents (e.g., bots, autonomous tools leveraging large-language-models) become active users of digital services, the AX-perspective becomes a key differentiator for platforms and apps.
Why does it matter?
When agents act on behalf of humans, products designed only for human users may create friction or inefficiencies when accessed by agents. Platforms that make it easy for agents to integrate and act will gain advantage.
Poor AX (e.g., unclear APIs, ambiguous interfaces for agents) can lead to failures, mis-use, or hallucinations in agent behavior.
The shift: from human-centric UX → developer-centric DX → now agent-centric AX.
Key Principles / Core Ideas
From multiple sources, the following are central to AX design:
Legibility & clarity: Agents need structured, machine-readable data, clear API semantics, unambiguous instructions.
Composability: Systems/tools should allow agents to chain steps, modularly build workflows, reuse components.
Feedback and recoverability: Agents need clear success/failure signals, error codes, retry paths.
Standards, interoperability & integration: For an ecosystem of agents, shared schemas and seamless integration matter.
Designing for agent + human collaboration: Recognizing that often humans and agents will co-work, so transitions/handoffs matter.
How it’s different / how to think about it
Unlike UX (which focuses on human perceptions, emotions, interface interactions), AX treats AI agents as a “persona” that uses the system.
Unlike DX (developer experience, focusing on APIs/tools for humans building software), AX focuses on how an agent (automated system) uses those tools.
It’s less about aesthetics and more about structure, semantics, modularity, machine-readability.
Implications for product and platform design
Products should consider agents as a user segment: e.g., can the agent easily discover capabilities, invoke tools, verify results?
Documentation and APIs may need agent-friendly versions: e.g., metadata, schemas, labeled I/O definitions, rather than only human-oriented docs.
Platforms that enable agents to “do more” (deploy infra, build workflows, integrate across systems) will capture value. Example: Netlify built an integration where a GPT agent deploys a website in one click.
Metrics for AX: track agent task completion rate, tool-usage accuracy, error recovery, plan quality.
Why now?
We are at a point where AI agents are no longer purely experimental—they’re being built into workflows, platforms, and consumer/enterprise systems.
The web and digital ecosystem are evolving: as agents do more of the “work” on behalf of humans (searching, choosing, executing), the traditional human-only UI/UX won’t suffice.
In short, “Agent Experience (AX)” is a design discipline that treats AI agents as first-class users of software. Products that want to thrive in the AI era will not only optimise for human users and developers, but also for the agents acting on behalf of humans. This means adopting machine-readable tools, clear semantics, reliable feedback loops, and seamless integration — ensuring agents can operate efficiently, reliably, and safely.
Source:
Agent Experience - https://agentexperience.ax
https://biilmann.blog/articles/ax-in-practice/
Join the conversation