Nielsen Norman Group: UX-context design

What the article is about

Published in July 2026, this piece by Tony Alicea at Nielsen Norman Group introduces the term “UX-context design” — defined as the practice of discovering and curating what an organization knows and wants into the context that guides everything its AI tools generate. The article is not speculative; it responds to a condition that already exists in many organizations: AI is generating interface work, but without organizational knowledge baked in, it produces average outputs rather than contextually appropriate ones.

Context: the shift from documents to context

The article builds on a structural observation: when AI models assist with interface generation, the output is only as good as the context they receive. A model without user research, design standards, or organizational terminology produces something generic. The same model with that information embedded in its context produces something that can actually be used.

Alicea points to Google Labs’ DESIGN.md as an early example of this approach — a machine-readable file that sits alongside code and contains design system values and guidelines. He proposes an expanded UX.md concept that would house research synthesis, interaction standards, glossaries, user models, and world models describing real-world usage conditions. The key shift is that these are not documents for human readers in a Confluence page or a shared drive. They are living resources intended to be pulled into AI context windows whenever interface work happens.

This matters because AI-assisted design is not confined to dedicated design tools. Developers use AI coding assistants to build interfaces. Product managers generate mockups with generative tools. The traditional role of the design team as gatekeeper of quality has given way to a distributed system where many people are generating interface work. UX-context design proposes a mechanism for maintaining quality standards across that distributed system rather than attempting to restore centralized control.

Key takeaway

Quality research and design standards must become continuously curated, readily accessible context that lives alongside product code — not archived documents that get referenced once and forgotten. The output of UX work is shifting from finished reports to maintained repositories that influence AI-generated output across the entire organization. This changes what UX practitioners produce and where they spend their time, but it extends their influence beyond what was possible when design review was the only enforcement mechanism.

Who it is useful for

UX designers and design leads rethinking what their deliverables look like when AI generates a substantial share of interface output. Product managers and developers building AI-assisted workflows who want to understand how design quality can be maintained without review bottlenecks. Accessibility practitioners will recognize the argument: machine-readable structure has always mattered, and this piece connects that existing investment to a new purpose.