WORK

ABOUT

LAB

Habitat

Cat furniture is bought blind. AR fixes that.

Cat furniture is bought
blind. AR fixes that.

Habitat

Cat furniture is bought blind. AR fixes that.

YEAR

2026

ROLE

End-to-End
Product Designer

TOOLS

Figma
Three.js
WebXR
Blender
Claude

STATUS

In Progress

Size mismatch

Size mismatch

Size mismatch

The #1 reason cat furniture gets returned,
solved before purchase

The #1 reason cat furniture
gets returned, solved before purchase

Zero friction

Zero friction

Zero
friction

WebAR in the browser. No app download, no QR per product

B2B2C potential

B2B2C potential

B2B2C potential

Embeddable by any pet furniture retailer
into their e-commerce

New B2B contracts
secured through
design-driven credibility

The problem

Cat furniture is bought blind. Owners research on Pinterest, choose by the photo, buy, and the product arrives the wrong size, or the shelves get installed in the wrong spot and the wall has to be redone.

The users

Cat owners in small urban apartments making high-consideration purchases: people who care about both their cat's wellbeing and their home's aesthetics, and who have been burned by misleading product photos before.

The problem

Cat furniture is bought blind. Owners research on Pinterest, choose by the photo, buy, and the product arrives the wrong size, or the shelves get installed in the wrong spot and the wall has to be redone.

The users

Enterprise managers in highly regulated industries: financial services, healthcare, retail. Decision-makers who needed to trust the product before they'd trust the data it showed them.

The problem

Cat furniture is bought blind. Owners research on Pinterest, choose by the photo, buy, and the product arrives the wrong size, or the shelves get installed in the wrong spot and the wall has to be redone.

The users

Enterprise managers in highly regulated industries: financial services, healthcare, retail.
Decision-makers who needed to trust the product before they'd trust the data
it showed them.

Research findings table showing six cat furniture pain points, their frequency from very high to medium, and how Habitat addresses each one through AR visualization

Three pain points emerged consistently across reviews, forums, and specialist blogs, all pointing to the same root cause:
no spatial reference before committing.

My role

My role

Solo project. I owned everything from research to coded prototype. Research, competitive analysis, persona definition, information architecture, user flows, UI design, 3D modeling in Blender, and front-end development with Three.js and WebXR. Claude was used as a coding collaborator to accelerate the development layer.

Finding the real problem under the obvious one

Finding the real problem under the obvious one

I mapped the problem through desk research: e-commerce reviews, cat owner forums, and specialist blogs. Three consistent pain points emerged: misleading product scale in photos, chaotic wall planning without spatial reference, and the paralysis of not knowing where to place furniture before committing.

A competitive analysis confirmed the gap: no product addressed cat furniture + AR as a primary experience. The closest competitor, Catit AR, hides AR behind a static 3D view with no catalog navigation.

Competitive analysis comparing IKEA Place, Catit AR, and Habitat across seven criteria including catalog integration, AR as primary state, and access cost — Habitat proposal meets all criteria

No existing solution combined an integrated catalog with AR as the primary state, and none focused on cat furniture specifically.

The user

45m² apartment

First cat

Lives with partner

"My apartment is 45m². Every centimeter matters. I want to give my cat a better quality of life without turning the living room into a pet shop."

The solution was a layered hierarchy: Contacts → Groups → Policies → Rules → Templates. Each layer added specificity without requiring the user to understand
the layer below it. Configuration became progressive, not overwhelming.

Without habitat

Research

Research

Can't visualize in the apartment

Can't visualize
in the apartment

Negotiation

Negotiation

"It'll look
huge here"

"It'll look
huge here"

Deadlock

Deadlock

Decision postponed

Decision
postponed

Giving up

Giving up

Problem unresolved

Problem
unresolved

With habitat

Visualizes in AR

Visualizes in AR

Tests configurations

Tests
configurations

Convincing

Convincing

Shows AR
in real time

Shows AR
in real time

Purchase

Purchase

Buys with confidence

Buys with
confidence

Result

Result

Happy cat,
happy partner

Habitat doesn't just help individuals decide, it resolves shared decisions that would otherwise stall indefinitely.

The solution was a layered hierarchy: Contacts → Groups → Policies → Rules → Templates. Each layer added specificity without requiring the user to understand
the layer below it. Configuration became progressive, not overwhelming.

AR as the primary state, not a hidden feature

AR as the primary state, not a hidden feature

Most AR implementations bury the camera behind menus or treat it as a secondary mode. The design principle here was the opposite: the camera is on by default, the catalog opens over the AR view, and objects appear in the scene immediately on selection.

The platform's configuration model emerged from real client behavior, not assumptions. Managers needed granular control, but operational overload was
a real risk.

Habitat interaction flow diagram showing the full user journey from opening the link through camera permission, product selection, AR scene management, and sharing — color coded by interaction type

The full interaction flow, color-coded by type: AR actions, UI/permission states, decision points, capture and share.

This was a deliberate tradeoff. WebAR through Three.js + WebXR sacrifices some spatial precision compared to native AR, but gains something more important for this audience: zero friction. No app store, no installation, no QR per product. A link is enough.

The solution was a layered hierarchy: Contacts → Groups → Policies → Rules → Templates. Each layer added specificity without requiring the user to understand
the layer below it. Configuration became progressive, not overwhelming.

Designing the test before the prototype exists

Designing the test before the prototype exists

A remote unmoderated usability test was designed to validate the core flows without researcher bias. Three tasks cover the full critical path: placing a product in AR, adding a second object without leaving the scene, and sharing the result via the native share sheet.

The platform's configuration model emerged from real client behavior, not assumptions. Managers needed granular control, but operational overload was a real risk.

Testing in the participant's own home, on their own phone — the only context where spatial AR actually makes sense.

The solution was a layered hierarchy: Contacts → Groups → Policies → Rules → Templates. Each layer added specificity without requiring the user to understand
the layer below it. Configuration became progressive, not overwhelming.

Result

Result

The usability tests are still ahead. But the project already validates something meaningful: a solo designer can go from discovery to coded WebAR prototype, with real-scale 3D models, an integrated catalog, and multi-object visualization, without a development team.

The usability tests are still ahead. But the project already validates something meaningful: a solo designer can go from discovery to coded WebAR prototype, with real-scale 3D models, an integrated catalog, and multi-object visualization, without a development team.

The B2B2C implication is real: any pet furniture retailer could embed this directly into their e-commerce. For them, it means reduced returns and higher purchase confidence. For the buyer, it means never buying blind again.

The B2B2C implication is real: any pet furniture retailer could embed this directly into their e-commerce. For them, it means reduced returns and higher purchase confidence. For the buyer, it means never buying blind again.

Reflection

Reflection

Designing for spatial interfaces requires a different mental model. You're not designing screens, you're designing how objects exist in someone's physical space. Every UI decision had to account for the fact that the user's environment is the canvas.

I would have recruited cat owners for generative research earlier, before defining the personas. The desk research was solid, but direct conversations would have surfaced edge cases I didn't anticipate, like renters who can't drill at all, or people with partners who disagree on aesthetics.

Run the usability tests, iterate on the friction points that emerge, and complete the 3D model library. The natural next step would be building a retailer-facing embed version to validate the B2B2C hypothesis with a real commercial partner.

← Back to Work

← Work

Zapper →

Let's chat! ↗
Let's connect! ↗
Let's chat! ↗
Let's connect! ↗
Let's chat! ↗
Let's connect! ↗