Rooted Academy
Spatial mentorship for environmental wellness — a photo-based room audit and design-mentorship app that explains the why behind why a home feels off, and teaches people to fix it themselves.
Project background
This crisis has shifted the conversation toward Environmental Health, as research proves our homes act as an "external nervous system." The colors, textures, and spatial layouts we live with directly impact our cortisol levels and mood. For the millions of people now spending the majority of their day indoors — remote workers, students — a visually "loud" or poorly arranged home can lead to chronic sensory fatigue.
The Problem
Users intuitively sense when a space feels "off" but lack the design literacy to diagnose why. With professional help inaccessible, many turn to superficial social media trends that offer quick fixes but fail to improve their well-being. There is a clear need for accessible, low-pressure tools that teach the principles of restorative design.
The Idea
Rooted Academy — a photo-based room audit, bite-sized design lessons, and optional non-commercial product suggestions. That shape came from the brief; whether it would actually earn people's trust is what research was for.
Why I chose this brief
My project utilizes the home as a fundamental and universal mental health tool. Synthesizing my background in architecture and human-centered experience, as well as my deep interest in psychology, I aim to leverage UX/UI design to make restorative design principles accessible to a broad audience, empowering users to modify their living spaces to better align with their sensory requirements.
Starting goals
User Goal
Help someone identify what's stressing them about their space and build design literacy to stop guessing.
Business Goal
Establish Rooted Academy as a trusted home-wellness mentor, earning revenue through ethical, contextual suggestions — not a hard sell.
Approach
Iterative, not linear — every phase fed back into the one before it:
Research
Interviews
- Six one-on-one, semi-structured interviews
- Remote via Google Meet, transcribed with Gemini
- Mixed housing — renters, a homeowner, living with parents, a full-time digital nomad
- Spanned the US and India, to see how housing control shapes people's relationship to space
Riskiest Assumption
"People value understanding the 'why' behind a suggestion, not just receiving a directive." If that turned out to be a design-literate bias rather than universal, the whole premise would need rethinking.
What surprised me
| Finding | Detail |
|---|---|
| Universal impact | Every participant tied their space to mood, focus, or relationships — one traced arguments with her husband to a wall she'd painted deep red. |
| Light beat everything | 5 of 6 named lighting, unprompted, as a mood/focus factor. |
| Regret was common | 5 of 6 described a purchase or change that didn't work as expected. |
| Reasoning won | All six wanted the why, not just a directive — regardless of design background. |
| Commerce diverged | 4 wanted product suggestions optional; 1 wanted automatic; 1 warned of "a confusing mess" from too broad a scope. |
"I traced arguments with my husband to a wall I'd painted deep red."
"Blue walls drove me crazy until I repainted it — I realized it reminded me of some sour memories from our childhood home."
Competitive landscape
A SWOT pass against DecorMatters, Houzz, IKEA Kreativ, Planner 5D, and DecAI found the same gap everywhere: each generates a new design or sells a catalog, but none diagnose the room you already have, and none explain their reasoning. DecAI drew consistent complaints about generic, boilerplate output — exactly the trust problem this needed to solve for.
| DecorMatters | Houzz | IKEA Kreativ | Planner 5D | DecAI | |
|---|---|---|---|---|---|
| Top strength | Gamified AI + AR redecorating, 10M+ users | 25M+ photos & 3M+ pros — a one-stop ecosystem | Free AI room scan, real IKEA products | True 2D/3D precision + AI + AR/VR, 40M+ users | Instant single-photo redesign, #3 grossing in category |
| The gap | No “why” — a decorate/shop tool, not a design-literacy teacher | No structured “audit my room” feature — browsing, not diagnosis | No emotional-wellness or design-literacy layer — purely transactional | No reasoning — a drafting tool, not a mentor | No design education, no purchase path, frequent spatial “hallucinations” |
Two personas, from six voices
Affinity-mapped in FigJam twice — once by theme, once re-sorted into persona-building buckets — to let two distinct people emerge cleanly rather than one muddy composite.
Wants the “why”
Reasoning behind every move, not just the move — a recurring ask across nearly every participant.
Trust & credibility
Skeptical of generic AI; real examples, before/afters, and professional input carry more weight than polish.
Constraints shape everything
Budget, rented space, and housing control (parents, landlords) surfaced again and again as hard limits on any suggestion.
Aesthetic & style is personal
Density, color story, and materials split by individual — no single “good” room exists.
Trust breaks fast
One bad AI read, or one generic recommendation, was enough to lose confidence in the whole tool.
A layer these personas add that my other capstones don't
Each card below carries its own Current User Behavior Journey Map, baked into the visual, tracing how that person actually moves through deciding on and acting on a change to their space today.
Define
Only after the interviews and personas did the real problem statements surface — not something I walked in with.
Problem 1 — The Self-Guided Beginner
Someone newly motivated to improve their home leans on trends and purchases as a stand-in for real judgment — without understanding the specific conditions in their own room (light, airflow, materials) that generic advice can't account for.
HMW help them diagnose what's specifically not working in their space — and why — before they default to a trend or purchase?
Problem 2 — The Design Enthusiast
Someone with some design literacy has been burned before by generic, one-size-fits-all advice that didn't account for their room's actual conditions, and is skeptical a new recommendation could be any different.
HMW build justified trust in a room-specific recommendation from the very first interaction?
Feature roadmap (MoSCoW)
Must-Have
- Photo Audit
- Reasoning-First Overlays
- Environmental Breakdown
- Lesson Cards
- Diagnosis-Based Suggestions
Should-Have
- Photo Privacy Controls
- Optional Suggestions + Budget Filter
- Before/After Preview
- Two-Way AI Chat
- Audit History
Won't-Have
- In-App Checkout
- Designer Marketplace
- 3D Room Builder
Each column maps directly to research: Must-Haves answer the problem statements directly; Should-Haves answer specific participant requests; Won't-Haves were cut deliberately to avoid the feature sprawl one participant warned about.
Information architecture
Sitemap and user flow, built before any wireframing, to map out every screen the app needed and how a person actually moves between them — audit, findings, lessons, history — before any visual design started.
Four task flows come out of that map, and they are the ones carried into both rounds of testing:
- Audit + Learn — photograph a room → receive the reasoning-first overlay → read the environmental breakdown → open the lesson card behind a finding. The core loop: every diagnosis has to be explainable, not just scored.
- Next Steps — from a finding → diagnosis-based suggestions → filter by budget → before/after preview. Where the app turns a critique into something actionable.
- Ask a Follow-up + Finish Audit — two-way AI chat on a specific finding → resolve or dismiss → close the audit into history. The exit path, so an audit has an end rather than trailing off.
- Learn a Topic — enter from the lessons library rather than from a photo, for someone browsing without a room to fix.
Mapping all four before wireframing is what kept the lesson content from becoming a separate section bolted on beside the audit — each flow re-enters the same lesson cards from a different direction.
Brand
Five values, pulled from research: Grounded, Insightful, Honest, Attuned, Nurturing. Japandi-restrained — the calm comes from restraint, not emptiness, since Sasha specifically craves negative space.
Color palette
Severity markers kept separate from the brand palette so diagnosis never reads as decoration:
Typography
Logo & inspiration
The mark is a jali pattern, referring to attunement with home and roots. The slanted “Rooted” comes from the organic lean of bamboo and bonsai — grounded, nurturing. The lighter “Academy” behind it stands for knowledge and insight rooted in everyday life.
Wireframes
Sketched every screen in a mockup app first — some questions still open ("PAGE or CARD?" noted next to the Lesson screen) — then built clickable mid-fi in Figma and tested via Figma Make before any visual design work began.
1. Home — start a new audit or continue the last one
2. Scan your room
3. Audit results — default view, 3 findings
4. Audit results — findings list expanded
5. Finding detail — Priority issue (overhead lighting)
6. Finding detail — full, with suggestions for improvement
7. Lamp suggestions
8. Audit History
9. Learn — search a topic
10. Learn — search results
Mid-fidelity wireframes — tested with 7 participants
What mid-fi testing changed: the “Show different lessons” action read as untrustworthy randomization, numbered suggestion/lesson lists were misread as sequential steps, and one participant missed “Continue Last Audit” entirely on first tap. Those findings are what drove the color, labeling, and layout choices carried into high-fidelity below — full results are in the Testing section.
High fidelity
Rebuilt every screen at full fidelity across four tasks: Audit + Learn, Next Steps, Ask a Follow-up + Finish Audit, Learn a Topic. The always-visible "Ask a follow-up" input answers Rohini's need for depth; tap-to-reveal reasoning on every finding answers the single most validated research insight.
1. Rooted Academy — splash
2. Welcome back, Jane! (home)
3. Scan your room
4. Your audit results
5. Finding detail — Priority issue (overhead lighting)
6. Lesson — The role of layered lighting
7. Finding detail — full, with suggestions for improvement
8. Lamp suggestions
9. Audit History
10. Learn — search a topic
11. Learn — search results
12. Lesson — Fundamentals of lighting in design
High-fidelity screens — first pass, before usability testing
Usability testing
Two moderated, remote, think-aloud rounds — mid-fi before visual design, hi-fi after — to separate structural issues from ones visual polish alone could fix. Tested against the same task flows mapped in the user flow: Audit + Learn, Next Steps, Ask a Follow-up + Finish Audit, Learn a Topic.
| Metric | Mid-Fi (7 participants) | Hi-Fi (5 participants) |
|---|---|---|
| Flows tested | 5 task flows | 4 consolidated task flows |
| Completion | 100% (all 5 flows) | 100% (all 4 flows) |
| Unaided completion | Passed target on all 5 | Missed on "Learn a Topic" — 3/5 (60% vs. 80%) |
| Correct first tap | Passed target on all 5 | Passed target on all 4 |
| Avg. SEQ (ease) | 4.3 – 4.9 across flows | 4.6 – 5.0 across flows |
| Biggest complaint | "Show different lessons" felt random (5/7) | Suggestions missed when collapsed (3/5) |
| What held up | Audit → findings flow intuitive for all 7 | Color-coded severity praised unprompted |
| Duration targets | 2 missed — narration/prompt confusion, not a usability issue | — |
| Numbered-list confusion | 5/7 misread suggestion/lesson lists as sequential steps | Fix held — dropped to 1/5 |
| "Continue Last Audit" | 1 missed it — first-tap failure, fixed by color not layout | Fix held — dropped to zero |
| Comprehension gap | — | 3/5 could use "Finish Audit" but couldn't explain what it does — discoverable, not comprehensible |
| Also praised | Suggestions felt optional, not pushy (6/7) | Optional-feeling suggestions (5/5), budget filter fit, before/after comparison |
Iteration — screen by screen
Each flagged screen became its own before/after pair, with the reasoning written down so nothing changed on instinct. A smaller 5-participant before/after survey confirmed these read as real improvements, not just changes.
Audit Results
Finding Detail
Product List
Learn — Lesson Search
Lesson Card
Full circle
Back to the three goals research set out at the start — shown here in their final, iterated form, not the first draft.
Diagnose, Learn, Act — validated
The core loop, carried all the way through iteration: diagnose what’s wrong, learn why it matters, act on a fix.
Learning, more than one way
Design and environmental-health knowledge doesn’t land the same way for everyone — so Learn supports browsing by topic chip, searching directly, or following a recommendation surfaced from the user’s own audit findings, all pointed at the same lesson library.
Final results & reflection
- Validated: real appetite for a product that explains the why behind design, not just a redesign or a shopping cart.
- Biggest challenge: the "learn first, then act" mental model was unfamiliar at first — this product needs stronger onboarding than I scoped.
- What I learned: the Japandi-calm aesthetic was right, but people wanted more visual life — video, warm color pops. I also stopped seeing onboarding as friction and started seeing it as infrastructure.
- What I'd change: prioritize onboarding from day one; build micro-interaction animation during hi-fi, not after.
- Next steps: a real onboarding flow, a short personalization step, closing the micro-interaction gap.
I'm most proud of delivering an MVP that met its objectives, and taking the initiative to run follow-up surveys validating my own design iterations.
My architecture and civil-engineering research background is what let me treat this capstone as a research problem first — and how I found the real gap: the missing why behind design decisions, and its direct line to environmental wellness.