AI agents are being asked to do increasingly complex work. Tasks can include using the web to get up-to-date information and even controlling your computer.
Agents have become daily active users of websites and products that were designed for people.
AX Design for AI Agents
Agent Experience (AX) is a new specialized discipline of UX, and brings new product design and development challenges.
It also presents a distinct opportunity for first-movers to get ahead of competitors.
AI agents are new primary users of digital products, so how do you start building and designing UI for AI users?
For a deeper guide to AX, visit flocker.md
Examples of Good AX Design
Connecting agents to Microsoft Teams and Figma are two effective UX×AX design examples, and are already being used by product teams. I go into more detail on agent interactions and AI agent tool design in the full article here.
The example from Figma shows an AI agent (Codex, OpenAI) creating a Figma canvas from a working code prototype, started with a simple prompt: “recreate this in my figma canvas”.
With a well-designed MCP tool the Figma team have reduced the barrier to entry for one of their core products — opening the door to faster development cycles and better collaboration between designers and engineers.
Definition of AI Agent Experience and AX Design
The term Agent Experience (AX) was introduced in 2025 by Netlify CEO, Matt Biilmann. Biilmann proposes that user experience (UX) design principles should now be extended to AI agents.
AX design is the specialized discipline of creating interfaces that AI agents can use efficiently and predictably.
Biilmann cautions us to “start consciously designing the AX of their products, or risk being replaced”.
A 2026 study of 856 tools across 103 MCP servers found that improving tool descriptions raised agent task success by a median of 5.85 percentage points and partial goal completion by 15.12%.
How to Improve Agent Performance with AX Design
For those looking to get ahead of the curve, I’ve written a practical guide to AI Agent Experience for work, UX designers, developers and teams, available on flocker.md.
The full article continues to explore:
- how product teams can start improving AX
- the relationship between AX and UX
- resources to start designing agent-ready websites and tools
- how skills, plugins, and MCP tools fit in
- how markdown, frontmatter, llms.txt and agent-accessibility improve ai performance
- semantics, tool names and tips to get started
If you are building products, websites, documents, or tools that AI agents could use, there will never be a time better time to get ahead with AI agent experience.
If you found this helpful and would like to see more articles likes this, share and make sure your team doesn’t fall behind.
Top comments (0)