The new elements of UX
In "The Elements of User Experience", Jesse Garrett laid out 5 planes of UX design. Building from Strategy at the bottom layer up to Scope, Structure, Skeleton, and Surface as the final plane. In the second edition, he augmented the planes to discuss “product as information” and included content requirements, information architecture and information design. While it works well for consumer websites, and it major contribution was to help clarified how design could engage throughout the development lifecycle in order to help ensure the success of the product.
However, I never felt this model captured the complexity of designing enterprise solutions. So I expanded these five layers to seven in order to provide more clarity around systems that require compliance and governance. And to take into account the variability that comes when platforms are used in multiple industries each with different user roles, each with their distinct needs, using the same system to accomplish different tasks.
Specifically it failed to address things like the multiple roles common to enterprise software. Like the old saying “it takes a village…” enterprise software needs to accommodate the needs of everyone in the organization, adapting to different job functions, and subsequently different workflows. Enterprise systems must support multi-user environments: admins, approvers, analysts, executives, and more.
Along with that is the importance of terminology and instructions. Information architecture doesn’t go deep enough. This is not just labels, tooltips, error messages, etc. But also domain specific terminology, onboarding copy. instructional text etc. that has to be designed with the users’ learning curves in mind—what is initially helpful quickly becomes clutter. And given the complexity of the workflows, and trying to create self-evident designs is naive. Consistency gives way to predictability, and warnings & confirmations are frequently seen as inconveniences.
For a more complete description, check out my previous article Planes of UX.
However with the arrival of Generative UX this model is out of date and need to be updated. There are too many fundamental shifts in user experience resulting from the introduction of the ephemeral technologies know as AI.
New planes of Generative UX
GenAI is already pushed aside the assumption that the “surface” is visual. While its true people can read faster than they can listen, nevertheless the Graphic User Interface we have taken for granted for 50 years need to be reconsidered. Indeed all the planes of the user experience need to be reconsidered with the inclusion of AI.
Surfaces, Spatial, Auditory, Tactile: How the system manifests
This layer includes all sensory interfaces: visual to be sure, but also augmentations in spatial environments, voice/audio cues, haptics, and micro-interactions. As well as context, the set and setting if you will for the interactions with the system. In GenAI the idea of an “interface” is no longer fixed; it can emerge dynamically based on a combination of the user’s input either explicit, such as a a prompt, or implicit such as action, or their context and modality. When well designed these new interfaces will allow the user to move seamlessly between them, indeed frequently taking advantage of multiple interfaces at the same time. Enterprise applications must now account for these multi-modal touchpoints across physical settings, devices, roles, and environments. As well as enable hands-free, situational interactions (e.g., voice commands in field work, gestures in manufacturing, etc). Or in cases where the enterprise is relying on a set of trusted orchestrated agents who simply perform their tasks, no interface at all save for the outcomes they produce and the audit capabilities to confirm the checks and balances.
Context & Conversational; How understanding unfolds
Terminology and instruction was not part of the original Elements of User Experience. But in enterprise solutions, they are critical elements to ensure the success of the design. These historically have included both domain specific terminology as well as contextual instructions required to various use cases to ensure compliance. However in these traditional systems these would be at most part of a library, but with AI these will no longer be static elements in the user experience. This layer is the heart of situational awareness and dialog-based interaction. In order provide the right level of context and instruction enterprise software needs to include domain-specific nuance, regulatory context, real-time governance, and interdependencies between tasks and roles. Systems must track and interpret who the user is, what they’re trying to do; what came before, what they are doing at the moment, and anticipate what might come next. It blends intent recognition, memory, conversation management, while reflecting environmental signals, to make the experience feel fluid and intelligent. Moving outside the chat box, enterprise solutions will need to have dynamically created interfaces that include contextual terminology as well as bidirectional instructions to enable effective conversations between the operator and the system.
User Aspiration & Evolving Models: Who the user is becoming
The original model briefly touched on the idea of user segments and the importance of understanding demographic differences and variations in needs/interests based on each segment and it was placed in the Strategy plane. However, in enterprise development the concept of roles play more central role defining the user experience. In order to effectively decompose complex heterogenous workflows for both efficiency as well as compliance it is often a requirement to segments access to various information and functions based on job titles and areas of responsibility, certifications, etc. Most systems rely on Role Based Access Control to make managing and tracking who has access to what easier all around. Role-based access control will need to be extended to address multi-tenant, shared, and restricted usage of MCP tools. But this layer goes beyond RBAC and stretches in into the evolution of identity management, security and authentication. But that will be covered in more detail in the Inferred Intention & Expectation plane.
At the same more advanced organizations will use these capabilities to enable long-term development goals for their teams based on behavioral signals, and in response to demonstrated career aspirations. In enterprise AI systems, this allows for both adaptive learning environments as well as proactive enablement, career-aware nudging, and even role progression. The system isn't just responding—it's guiding, coaching, and growing with the user over time.
Generative Hierarchies: How structure adapts in real time
This reimagines traditional information architecture as dynamic and generative. Traditional IA while at times designed to be flexible or to take into account a set of pre-defined roles, with the introduction of AI information architecture will need to “float”, it will need to be dynamic and responsive to what is generated. AI powered systems create and prioritize informational hierarchies based on meta-data changes, generating different views, steps, and ontologies on-the-fly based on context provided by the data itself and the user's intent/prompt. The idea of roles also becomes dynamic. Given the intent of user and their organization's policies around governance—which themselves will evolve based on the classifications applied in situ to the generated output from the systems, roles will be personalized and highly contextual. Given the inflow of new data to an existing data set may generate new IP or at the very least new capabilities (think medication, code, financial predictions, etc.) the access to that information will need to be tied to something other than a "role" that was created before the new capabilities were created. For the enterprise, this is about AI-curated task flows and information layers—adjusting to complexity, urgency, and enterprise governance constraints.
Ever-Changing Capabilities: What the system can do today (and tomorrow)
In GenAI, the system’s skillset is not fixed. Capabilities evolve continually via architectural model updates, plugin ecosystems, learned patterns, and organizational prompts. Not to mention Retrieval-augmented generations using updated private information and intent recognition who’s refinements of the model’s ability to infer what the user really wants is accelerated in the context of enterprise workflows where you may have hundreds of users repeating the same patterns every day. (Including human-in-the-loop actions to ensure oversight, accountability and ethics.) What this means from a UX perspective is that predefined workflows are now legacy; emergent workflows will the more common approach; adapting to both the collective behavior of teams as well as responding the descriptions, requests and prompts. UX design will be focused on prompt curation and defining parameters for the resulting generation.
This layer also ensures that the system can explain, expose, and safely extend what it can currently do—while managing change, risk, and transparency. In enterprise solutions, it’s essential to balance innovation with control, versioning, and compliance. Placed between the User Aspiration and Inferred Intention these capabilities are where the enterprise’s data is transformed into value. And as such it is important to ensure the organization has the means to protect and effectively monetize these assets.
Inferred Intention & Expectations: What the user means
This is the predictive core of GenAI UX—what the system infers the user is trying to achieve. It blends behavioral signals, prompt patterns, past actions, and role-based heuristics. The experience becomes anticipatory: accelerating likely workflows, auto-generating starting points, or alerting to blockers before they arise. It must gracefully handle ambiguity, edge cases, and corrections—building trust without overstepping. This is where role based access becomes the focus. Traditionally RBAC governs static screens, functionality, features, and data sets. However with the emergence ofAI-driven, generative UX environments, that will all be generated in real-time, breaking the assumptions RBAC relies on. It will need to focus on the intent and what the user is trying to do rather than on which button they click. As part of the data strategy it will be necessary to move to an Attribute-Based Access Control (ABAC) model that uses the metadata tied to both the underlying data being accessed as well as the user’s attributes. Alternatively, a Policy-Based Access Control (PBAC), or just-in-time entitlements, could be used to grant access on a per sessions basis to data and capabilities based on request type, risk level, and context. (Where again context could include the physical setting as well as the other people working proximity to you)
Data Strategy: What fuels the intelligence
The foundational layer, encompasses how data is collected, structured, governed, and used. It includes data quality, permissions, lineage, feedback loops, fine-tuning sources, and model alignment. In enterprise, data strategy is tightly bound to security, explainability, vertical compliance, and the ability to adapt GenAI to organization-specific language, workflows, and success metrics. Moving forward instead of optimizing data for human consumption or static workflows, enterprises must now structure and govern their data based on the needs/requirements of agents operating in generative, adaptive contexts. This means shifting from a warehousing model to a real-time, contextually availability, included semantic layering, and vectorized indexes. Augmenting human-readable meta-data with machine optimized metadata such as embeddings, entity relationships, and intent. To accomplish this shift organizations will need to move to composable integrations, and from monolithic pipelines to modular agentic integrations focusing on the right data, fast with confidence or relevance scoring and most importantly traceability. To maintain these systems, dashboards will give way to natural summarizations and action recommendations—which themselves could be passed to agents to resolve. Most importantly to achieving a successful strategy will be the removal of the silos that currently permeate enterprise. The new version of Conway’s Law will be if you can identify a companies org chart by their data strategy and its manifestation in their generative UX.
While foundational for design’s role in development, Garrett’s The Elements of User Experience five planes for user experience is now out of date. With the arrival of GenAI we need to challenge the primacy of the graphical interface with the introduction context-aware, multimodal, and adaptive experiences. It’s time to rethink UX as an emergent, AI-mediated systems. Like a building on a floating foundation, generative user experiences achieve balance—not by resisting change, but by responding to it. They rest not on fixed inputs, but on the dynamic equilibrium of evolving data, adjusting in real time to preserve coherence and relevance not a static hierarchy of layers.
Garett’s Five Layers
Below is a brief description of the five planes from Garrett’s Elements of User Experience.
Surface — How the product looks and feels. This layer focuses on visual elements like typography, color, and imagery. Its purpose is to ensure the product is aesthetically appealing and easy to understand.
Skeleton — How users interact with and navigate the product. This layer focuses on interface elements such as buttons, menus, and navigation systems. Its purpose is to optimize usability and streamline user flow.
Structure — How features and content are organized. This layer defines the underlying framework of the product, arranging information and functionality into a coherent system. Its purpose is to support intuitive interaction and information access.
Scope — What the product includes—and excludes. This layer translates strategic objectives into specific features and content. Its purpose is to define the product’s functional boundaries and capabilities.
Strategy — Why the product exists. This foundational layer defines the product’s goals by aligning business objectives with user needs. Its purpose is to establish a clear vision for success.
