Designing for Composability: Rethinking UX Design Beyond the Application Boundary

Anticipating AI as the Orchestrator

The rise of agentic AI accelerates this shift toward reuse. McKinsey’s 2025 global AI survey found that 88% of respondents’ organizations regularly use AI in at least one business function, while 23% are already scaling agentic AI somewhere within the enterprise and another 39% have begun experimenting with AI agents. In a 2026 article on scaling agentic AI, McKinsey noted that nearly two-thirds of enterprises have experimented with agents, but fewer than 10% have scaled them to deliver tangible value, often because of data limitations.

These findings matter for UX design because agentic AI doesn’t merely change how people interact with a screen. It changes how workflows get assembled. An AI agent might pull from several systems, infer a user’s intent, select a tool, and trigger an action, then summarize the outcome. The user experience becomes less like navigating an application and more like supervising an orchestration layer.

While this doesn’t make UX design less important, it does make UX design more difficult to locate. If an AI agent becomes the orchestrator, UX designers must increasingly shape

• tool discoverability
• data context
• permission boundaries
• confidence signaling
• escalation paths
• exception handling
• auditability

In other words, we must design what the agent can understand, what the user can inspect, and where the system must slow down.

Avoiding Composable Confusion

Composability offers flexibility, but flexibility can quickly become confusion. When teams decompose products into components, APIs, and reusable services, they might gain architectural agility but weaken experiential coherence. A status label that made sense in one product might be ambiguous in another. A data value might be accurate but also lack freshness, provenance, or supporting assumptions. A recommendation might be valid only within certain constraints that disappear when another system uses that recommendation. UX designers can contribute meaningfully to solving such problems.

Composable experiences require semantic discipline. UX designers should help define not just what information appears, but what that information means, where it came from, how current it is, and what decisions it should or should not support.

This changing role for UX designers is especially important within enterprise environments, where users often make decisions that carry operational, financial, safety, or compliance implications. In such contexts, reuse without context could be dangerous.

Reuters has reported Gartner’s prediction that more than 40% of agentic-AI projects might be canceled by the end of 2027 because of rising costs, unclear business value, and immature agentic capabilities. However, the same report noted Gartner’s prediction that 33% of enterprise-software applications could include agentic AI by 2028.

We must consider these two predictions together. While agentic capabilities are likely coming to enterprise software, many of these efforts will fail unless organizations first establish their clear value, governance, and contextual integrity. Composable UX is one way to reduce that risk.

Shifting UX Upstream

When we’re assembling user experiences from modular capabilities, UX design must move upstream. All too often, teams bring in designers after architectural decisions have already shaped what’s possible. (Any UX designer working in a large enterprise can likely relate to this.) When this occurs, teams might expect UX designers to make disjointed capabilities feel cohesive through surface-level user-interface design. This is insufficient for composable ecosystems. UX designers must influence the following:

• information architecture
• data models
• API contracts
• event structures
• permission models
• system states
• notification logic
• error taxonomies

While this doesn’t mean every UX designer needs to become a software architect, we do need to become more fluent in the design consequences of software architecture.

In my past UXmatters columns, I’ve often suggested that UX professionals must grow beyond narrow interpretations of their roles. My cloud-skills series made a similar point: focusing too heavily on roles and positions can skew thinking toward staffing before teams understand the value of a product to users and the business outcomes they must deliver. Composability raises the stakes for this concept. If UX designers remain focused only on designing screens, we’ll join projects when it’s too late to shape the experience.

Establishing a Composable UX Design Framework

Designing for composability requires a different lens. The following principles can help establish a framework for composable UX design:

• Design capabilities as portable meaning. A reusable capability shouldn’t depend entirely on its native user interface to ensure it makes sense to users. Its outputs should carry adequate context with them, including their source, timestamp, confidence level, assumptions, constraints, and recommended next steps. After all, a value without context isn’t an insight, and a recommendation without boundaries isn’t guidance.
• Separate observations, interpretations, and actions. Composable systems often blend raw data, inferred meaning, and recommended actions. This can create confusion, especially when AI is generating the recommendations. UX designers should help distinguish what the system has observed, what it has inferred, what it recommends, and what it has already done.
• Design for unknown consumers. We can no longer assume that only the native application will consume a product’s outputs. Other tools, dashboards, agents, or workflows might reuse them. Designing for unknown consumers requires making meaning explicit and reducing hidden dependencies. It means asking, “Would this still make sense if it appeared somewhere else?”
• Make constraints visible. Composable systems can encourage unintended combinations of components. Users and agents might compare things that are not comparable, combine data from mismatched time frames, or infer causality where none exists. Therefore, UX designers must help expose constraints, not bury them in design documentation. Constraints are an important part of the user experience.
• Preserve human judgment points. Composition should not eliminate human oversight. Instead, it should clarify where human judgment matters most. Human oversight should connect directly to the judgment layer. If intelligent systems assemble workflows dynamically, we must identify the points at which users should review, approve, override, or investigate the system’s outputs.

Reshaping Product Strategy

Composable UX design is not just a design concern; it’s a product-strategy concern. If product teams continue optimizing only for engagement within the boundaries of the native application, they might miss where value is actually moving. In some cases, the most valuable product might not be the one users log into most often. It might be the one whose capabilities users trust most, are most reusable, and which designers have intelligently orchestrated across an ecosystem. This requires a different definition of product success. What might this look like? Instead of asking only, “How do we keep users in our application?” teams might ask the following questions:

• How reliably can others build upon a product’s capabilities?
• How clearly do the product’s outputs communicate context?
• How safely can AI orchestrate the product’s workflows?
• How well does the product support judgment beyond the native user interface?

Asking these questions might feel uncomfortable because they challenge familiar mental models. Product teams often want to own the destination of a user journey. But in composable ecosystems, value might come from being an indispensable participant in a broader workflow.

Conclusion

Composability doesn’t make UX design disappear. Rather, it relocates the work to another, more strategic point in the design process. Instead of focusing only on the screens and flows within a bounded application, UX designers must increasingly shape the meaning, constraints, and decision points that allow capabilities to travel safely across systems. We must think more deeply about architecture, data, interoperability, and AI orchestration—not as purely technical concerns, but as user experience concerns.

Which application has the most elegant shell probably won’t define who wins the competition to own the next generation of enterprise software. Instead, the determinative factor will likely be which products can participate most intelligently in ecosystems they do not fully control. UX designers now have a meaningful opportunity: helping organizations to design not just better user interfaces but better conditions for understanding, trust, and human judgment across increasingly composable systems.