Experience Design (UX/CX)
Experience Design at the Intersection of Human Need and Intelligent Systems
The arrival of AI in products creates a new design challenge: how do you design for a system whose outputs are probabilistic, whose reasoning is opaque, and whose capabilities change with every model update? The answer begins with the same discipline that has always produced great products — deep understanding of the people using them.
Discovery: The Work That Makes Everything Else Easier
Design discovery is not a phase with a start and end date. It is an ongoing practice of building and maintaining a clear understanding of who your users are, what they are trying to accomplish, and where the current experience fails them.
The best discovery work is contextual. Watching a user navigate your product in their actual work environment — distracted, time-pressured, with three browser tabs open and a Slack notification arriving — teaches you things a lab session never will. The goal is not to collect data about users; it is to develop genuine empathy with their experience.
For enterprise products specifically, discovery must account for multiple user archetypes: the power user who uses the product eight hours a day and has developed workarounds for its limitations; the occasional user who uses it once a week and cannot remember where anything is; and the administrator who never uses the core product but is responsible for its configuration and maintenance. Each requires a different design response.
Designing for AI: New Principles for Probabilistic Interfaces
Designing interfaces for AI-powered systems requires principles that did not exist in the era of deterministic software.
Calibrated confidence: AI outputs should communicate their confidence level — not as a raw probability score, but as a human-readable signal. "I'm not certain, but based on the last three months of data..." is more useful than either a false certainty or an unhelpful disclaimer.
Graceful uncertainty: when the system does not know, it should say so clearly and escalate gracefully to a human or a fallback. Uncertainty is not a failure; it is information. Design it as such.
Explainability at the right level: users do not need to understand how the neural network works. They need to understand why this specific output is being shown to them, and what they can do if it is wrong. "We suggested this because..." with a visible, editable rationale is more valuable than any model card.
Human override as a first-class pattern: every AI recommendation, classification, or action should have a visible, low-friction path for a human to override it. Users who trust the system use the override rarely; users who distrust it never use the AI at all. The override path is what builds trust over time.
Customer Experience Strategy: Connecting Touchpoints Into a Journey
Customer experience (CX) design operates at a level above individual screen design. It asks: across every touchpoint — digital product, support interaction, billing notice, onboarding call, renewal conversation — what is the customer's cumulative experience of our brand?
Journey mapping is the core tool of CX strategy. A well-constructed journey map shows: the customer's goal at each stage; the touchpoints they encounter; the emotions they experience; the friction they hit; and the organisational actors responsible for each interaction. It makes visible the gaps between what the organisation believes the experience to be and what customers actually experience.
The most valuable journey maps are built with customers, not about them. Running a co-design session where customers build their own journey map surfaces insight that no amount of analytics data can match. The map becomes a shared artefact that aligns product, marketing, support, and operations around a common understanding of the customer's reality.
Interaction Design: The Details That Make the Difference
Interaction design operates at the level of individual moments: the transition that tells you an action succeeded, the error message that tells you exactly what to fix, the empty state that guides you to your first meaningful action, the loading state that keeps you from feeling abandoned.
These moments are individually small. Collectively, they define whether a product feels effortful or effortless. Research consistently shows that users attribute quality to products that "just work" — meaning the micro-interactions at every step are coherent, predictable, and in service of the user's goal.
For complex enterprise products, interaction design must account for state complexity that consumer products rarely face: multi-step workflows with branching paths, collaborative editing by multiple users, data that updates in real time, and actions with consequences that cannot be easily undone. Designing these interactions well requires not just aesthetic judgment but a deep understanding of the underlying data model and business logic.
