From UX to AX: What AI agents are really changing for designers

From traditional UX to Agent Experience (AX), AI agents are redefining the rules of interface design. This analysis examines key developments, from the MCP protocol to concrete opportunities for UX/UI teams.

An article by Yasmina Belhadj, UI/UX Designer at UX-Republic.

Two maps, three years apart

You only need to compare two images to measure the speed of change. In 2023, a chart circulated on LinkedIn listing AI tools by category: Text, Video, Audio, Face, Legal, Voice, Meeting, Productivity, Coding . About ten boxes, a few logos per box. Claude was barely included. The term that dominated everywhere at the time: generative AI.

Three years later, the mapping of LLMs and AI systems resembles a star chart—dozens of bubbles organized by function ( Creative & Generative, Reasoning & Knowledge, Interaction & Assistance, Perception & Recognition, Ethical & Trustworthy AI… ). The number of players has exploded. And one term has crept into every product conversation: agentic AI.

This is not just a change in vocabulary: it is a fundamental shift in what we expect from an interface.

Generative AI vs. Agentic AI: The difference that changes everything for a designer

Generative AI creates content from a prompt (text, image, sound, video). Agentic AI , on the other hand, acts autonomously to achieve a goal: it decides, sequences steps, and adapts.

In both cases, the user is faced with a conversational interface where they express an intention rather than navigating through a traditional menu. But the agent goes further: it consults external data, interacts with applications, and accesses business documents.

Figma MCP Server Guide

Our interfaces, originally designed exclusively for humans, thus become the terrain of an agentic experience — we now speak of AX (Agent Experience) or Agentic Design.

The key role of the MCP (Model Context Protocol) protocol

What makes this evolution technically possible is the Model Context Protocol (MCP) : a standard that connects an AI model to the tools and data it needs to function. Thanks to the MCP, an agent can search for information in a CRM, consult internal documentation, or create a task in a management tool, without requiring specific integration for each service.

Designing for agents: what does the AX look like on a daily basis?

From UX to AX: Why Agent Experience is the Next Frontier in Business AI

The role of the UX Designer is no longer limited to optimizing a human user journey. It is now necessary to integrate the needs of an agent so that it can function effectively:

  • Structured and consistent data: an agent who has to fill out a form or retrieve information needs clearly named and tagged fields, not a graphical interface designed solely for the human eye.

  • Usable documentation: content that is readable by a human and interpretable by a machine (alternative texts, metadata, semantic hierarchy of content).

  • Standardized exchanges between systems: for example, Figma's MCP server allows an agent to directly extract a design system variable or component, rather than trying to guess a value from a screen export.

In concrete terms, a well-documented design system exposed via an MCP server makes all the difference between an agent capable of producing a screen perfectly aligned with the brand guidelines and an agent that reinvents its own color palette with each generation — the famous “clean but generic” design syndrome that many designers are now seeing.

UX remains human-centered, but user journeys are evolving.

The motto remains unchanged: human first , the user at the center. What is changing is that the user's needs are evolving along with the interfaces. The rise of chatbots and virtual assistants is reshaping where and how users access information.

A strong indicator of this shift: Google Overviews (AI-generated summaries at the top of search results) have had a measurable negative impact on the traffic of many websites. When AI answers directly on the search page, the user no longer clicks through to the site—and therefore no longer engages with the designer's interface. The traditional user journey we optimized yesterday (from search to site) loses some of its relevance when the AI ​​intercepts the user earlier.

State of the Designer 2026: Designers address the uncertainty of the customer journey | Figma Blog

This shift is confirmed by the State of the Designer 2026 study published by Figma: while the majority of designers surveyed state that AI tools help them work faster and create better designs, a significant proportion (26% strongly disagree, 49% disagree) also believe that these tools sometimes slow them down. The productivity gain is therefore neither automatic nor uniform.

The new skills of the designer in the AX era

Faced with these changes, a new range of skills is required for design teams:

  1. Understanding the agent's reasoning: knowing how an agent approaches a task and what data it consumes.

  2. Strengthen Tech & Data collaboration: work closely with Data teams and developers to structure information.

  3. Thinking about inter-system interactions: designing how different systems and agents interact with each other.

  4. Adopt a more technical vocabulary: to appropriate a lexical field at the crossroads of design, data and software architecture.

AI and UI: the 3 current use cases

In terms of user interface (UI) design, three main scenarios are emerging in the tools on the market (Figma Make, Lovable, Uizard, Galileo, Claude Design, Google Stitch…):

  • Scenario 1: Designing a product from scratch.

  • Scenario 2: Generate variants from an existing design.

  • Scenario 3: Generate screens from an established design system.

After feedback and tests on these tools (including image generation), Scenario 3 proves to be by far the most relevant for a designer in an agency: this is where the AI ​​tool respects an already established brand identity, instead of inventing a new one at each iteration.

https://trello.com/b/6XLTeqJn/boite-a-outils-designia

Conclusion: AX design as an extension of UX

We often read that AI speeds up the designer, increases production, or diminishes empathy. Yet, the center of gravity of UX remains the user. A new discipline simply builds upon this: AX Design . Far from replacing UX, it extends it to a new key stakeholder: the AI ​​agent.

Sources and further reading:

Yasmina Belhadj, UI/UX Designer at UX-Republic