A mixed reality tarot card reading app for HoloLens — grounded in market sizing, competitor analysis and three distinct user personas before a single 3D asset was built.
Tarot reading is centuries old, but access to a good reader isn't universal, and many people are curious without wanting the commitment of an in-person session. Virtual Tarot's premise: use mixed reality to create a private, personalized reading experience that feels like sitting across from a real reader — accessible to skeptics and believers alike.
Before designing anything, I sized the opportunity: the AR/VR market was valued at $27.6bn in 2021 and projected toward $856bn by 2031, while the global tarot card market itself was valued at $1.165bn and growing steadily — two converging trends supporting a mixed-reality tarot product as more than a novelty.
I then benchmarked four existing tarot apps (Galaxy Tarot, Labyrinthos Tarot Reading, Golden Thread Tarot, and Fool's Dog) across technology, features, pricing, platform, and — critically — real customer reviews, to find where competitors were actually falling short rather than guessing.
Educational content, wide deck selection, simple onboarding, and in one case, full VR immersion (Fool's Dog).
Most features were gated behind in-app purchases, and reviews consistently flagged shallow personalization as the core complaint.
A genuinely personalized, conversational reading experience — not just a scanned-card lookup — was the clear gap to design toward.
A structured SWOT kept the concept honest about its real risks, not just its appeal: a HoloLens-based experience is highly immersive and private, but the hardware cost and Windows-only technical complexity meaningfully narrow the addressable market — a constraint I carried into every later decision rather than designing around an idealized, unconstrained user.
Rather than one generic "user," I defined three personas with genuinely different reasons to try the same product — because a skeptic, a believer and someone looking for entertainment need different things from an identical feature set.
Rachel Patel
"I'm not sure I believe in the accuracy of tarot readings, but I'm curious to see what insights the cards might offer me."
Business professional who wants education and transparency about how the reading works before she'll trust any of it.
Sofia Singh
"I've always been drawn to spirituality and I'm excited to explore tarot readings as a tool for self-discovery."
Frequent, invested user who wants depth, personalization and an experience she can return to regularly.
A third persona, Tyler Lee — the "entertainment enthusiast" — wanted a fun, shareable experience with friends rather than personal insight, which pushed save/share functionality onto the feature list alongside the more introspective use cases.
I translated the personas into a shared use-case flow and specification table — including alternate flows for edge cases like no internet connection or fewer than three cards selected — then storyboarded the full session scene by scene, from an introductory welcome through card selection, AI-driven interpretation, and a closing summary with save/share options.
Core interaction flow — every step maps to a specific persona need identified earlier.
The working prototype was built in Unity for HoloLens using the Mixed Reality Toolkit, with 3D tarot card and environment assets modelled in Blender and a custom "Virtual Tarot Lady" character sourced from Mixamo and restyled to fit the experience. The standout technical decision was integrating OpenAI's ChatGPT directly into Unity via C#, so the virtual reader could generate genuinely personalized, conversational interpretations in response to a user's actual typed or spoken question — not a lookup table of stock meanings.
Interaction design leaned on intuitive hand gestures for card selection and placement, minimalist UI so the cards stayed the visual focus, and layered sound design (spatial audio, ambient music, responsive feedback) to reinforce immersion without overwhelming the experience.
Technical testing ran across 23 build iterations, covering functionality, navigation, interaction, sound and the OpenAI integration itself — the highest-risk technical component in the entire build. Every test case passed by the final build, including validating that the AI integration correctly handled both keyboard and voice input.
Virtual Tarot is the project where I most deliberately practiced product thinking before design thinking — market sizing and competitor gaps shaped the brief before personas shaped the screens, and personas shaped the screens before any 3D asset got built. It's also where I got hands-on with integrating a live AI model into an immersive prototype, which now underpins how I think about designing AI-enabled experiences responsibly in my current role.
Want the full research, wireframes, storyboards and prototype video for this project?