Clinicians had no good way to measure and elevate their procedural skill. The existing options fell short in two ways:

• Books and observation don’t transfer. Reading a technique, or watching another doctor perform it, is fundamentally different from doing it yourself. There was no safe, repeatable way to practise the actual motions of a procedure and get assessed on them.
• Plenty of training, almost no assessment. The market had training platforms, but very few that assessed, evaluated and scored skill-based learning and then fed that back into a personalised improvement loop.

Research backed the premise: immersive VR learning showed better real-world outcomes than traditional study. The opportunity was to build something more innovative and approachable — an experience a clinician could run from the comfort of home, without booking physical theatre or lab time, that got them as close to a real procedure as possible while removing the real-world risk.

The core insight: lower the risk-to-reward of learning. Let clinicians make mistakes in a place where mistakes are free.

I want to be precise about ownership, because this was a cross-functional project:

I personally owned:

• Brand identity — logo, colour system, type styles, and the overall visual language
• Design system — translating that identity into reusable UI: buttons and their states, navigation patterns, spacing/padding/gap systems
• Interaction & flows — user flows for moving forward and back through the experience, error prevention, and loading states
• Stakeholder communication — building and presenting concepts to stakeholders and the UCH clinicians, then iterating on their feedback in bi-weekly reviews

I collaborated on: the headset implementation, via a structured handover to a developer who embedded the system into the live experience.

I studied the two most relevant players — one direct, one aspirational benchmark — to find the gap we could own.

FundamentalVR (now FundamentalXR) — A capable product, but the brand felt unapproachable, overly commercial and clinical-to-a-fault. It read as something built by domain experts without a design lens. Takeaway: there was room to win on warmth and approachability.

Osso VR — The bigger, more established competitor, and an early mover onto Apple Vision Pro with a strong, distinctive colour identity (which we deliberately steered away from to avoid brand confusion). The experience itself looked excellent and was used globally — something we wanted to emulate in reach. But the surrounding product let it down: a laggy, desktop-first website that performed poorly on mobile. Takeaway: it functioned as a training system but felt out of touch across skill levels — not genuinely human-centric.

What this told us: the market had capable technology but a missing human-centred design layer. Our differentiator wouldn’t be “another VR trainer” — it would be the contemporary, approachable, genuinely user-friendly one that worked for clinicians at every level.

This is what kept the design honest. Rather than designing in a vacuum, I ran the work past practising gastroenterologists from UCH and let their domain expertise direct decisions I couldn’t have made alone.

Their feedback reshaped the spatial layout directly. On the placement of interface elements and tools, they’d correct assumptions with real procedural logic — “you actually wouldn’t place things here, you’d be picking it up from here.”That turned abstract UI positioning into something that mirrored how a clinician actually moves and reaches during a procedure.

The validation was just as telling:

“This is as close as we’ve come to experiencing an operating theatre.”

“This improves the risk-to-reward if something goes wrong.”

The brand had to do something specific: feel contemporary and approachable in a category that mostly felt cold and clinical. That positioning — directly informed by the competitor gap — drove the logo, colour and type decisions, which then had to survive translation into a 3D, head-mounted environment.

Designing a brand for a flat screen is one thing. Designing one that stays legible, consistent and on-brand inside a headset — at varying depths, sizes and viewing angles — is another. The identity wasn’t the deliverable; it was the foundation for a system that had to hold up in space.

This was the hardest and most interesting part of the project, and the part I’d want any UX team to ask me about.

Getting the spatial system right. Designing UI for a headset meant learning the correct spatial terminology and then holding button spacing, padding and gap systems consistent across a 3D environment — while accounting for transitions between different states. Consistency is hard enough on a 2D grid; in a headset, depth and head movement make it harder.

Designing for comfort over time. A real constraint was duration: how do you keep someone in a headset long enough to complete a meaningful assessment without them feeling sick? That pushed decisions about pacing, motion, visual stability and how much was on screen at once — comfort became a design requirement, not an afterthought.

Adapting to a changing interaction model. The experience began with gaze tracking, which dictated how buttons had to look — active/selected states needed to read clearly when the only input was where you were looking. Midway, the input model shifted to hand tracking, which was a genuinely difficult adaptation: the same interface now had to support reaching, grabbing and direct manipulation. Re-thinking button states and affordances for hands-instead-of-gaze was one of the most demanding iterations of the project.

The core user flow. I designed the journey so a clinician could: sign in to their portal → choose to take the full assessment or select specific modules → complete the experience → receive their results and an overall score. Throughout, the priorities were error prevention (making the wrong action hard to take), and clear, low-friction navigation forward and back.

Designing while advocating for design. Worth being honest about: I was sometimes working with team members who didn’t initially see design as valuable. Part of the job was demonstrating — through presentations and bi-weekly stakeholder reviews — why these decisions mattered to the end experience and the outcomes.

Deft MD soft-launched and is ongoing. The platform is described as a clinically-validated immersive assessment tool, and the underlying approach has been published in peer-reviewed research (GIE Journal, 2025).

Everyone who has used it so far has responded strongly — the clinician feedback above speaks to that. The next phase is funding a larger team to take it from a validated soft launch to a fully product-ready platform.

What I’d take into the next iteration: the gaze-to-hand-tracking transition proved how much interaction-model assumptions ripple through an entire interface. With more runway, I’d want to formalise the design system documentation for spatial components and run structured usability testing across a wider range of clinician skill levels — closing the “human-centric for all levels” gap I’d identified in the competitor analysis from the start.