Case Study · Social Impact · AI Product

AdaptAI

AI-designed adaptive clothing that treats disability as a design opportunity — not a constraint.

My Role
Founder & Researcher
Context
Cornell University
Timeline
Aug 2025 – May 2026
Tools
VLMs · Python · Figma
Visit live product ↗
100+
Students Supported
40%
Cost Reduction
28–35%
Artisan Income Growth
30+
Brand Studies Conducted
01
Overview

A quick read before you dive in

What it is

An AI-powered platform that generates personalized adaptive outfit designs with hidden accessibility features — magnetic closures, side openings, seated silhouettes — built into beautiful, trendy looks.

Why I built it

Fashion is a form of self-expression and dignity. Excluding people with disabilities is not just a product gap — it is a social justice issue. The goal: eliminate the barrier between disability and style.

Who it serves

Cornell students with disabilities (100+ served through SDS), wheelchair users, people with sensory sensitivities, mobility-impaired individuals, and the artisans who serve them.

My contribution

I led the end-to-end product: user research at Cornell SDS, pain-point prioritization, solution design, VLM/Python prototyping, and shipping the live web app.

02
Problem

Adaptive fashion is broken in four specific ways

The global adaptive fashion market is nearly invisible. Over 1 billion people with disabilities are routinely excluded from mainstream fashion. Standard clothing ignores wheelchair users, people with sensory sensitivities, limited dexterity, and unique body shapes. The 'accessible' options that exist are clinical, expensive, and strip people of their identity.

At Cornell's Student Disability Services, I saw this firsthand. Students who needed adaptive clothing either couldn't find it, couldn't afford it, or were forced to wear something that made them feel less — not more — like themselves.

Pain 01

High cost

Bespoke adaptive garments run $300–500+ per item, pricing out most students.

Pain 02

Limited stylish options

Choices are functional but rarely fashionable — no self-expression.

Pain 03

Poor accessibility

Standard clothing ignores mobility aids, sensory needs, and unique body shapes.

Pain 04

Clinical-looking designs

"Accessible" often means visibly medical — stripping users of identity.

03
User Research

10+ interviews. 30+ brand studies. One clear signal.

10+
User interviews
30+
Brand studies
Cornell SDS
Field partner

Through 10+ deep-dive interviews and 30+ brand studies, two distinct user groups emerged — each with urgent, unmet needs that mainstream fashion completely ignores.

Primary User

The Person With a Disability

Students, young adults, and individuals with physical disabilities, mobility challenges, or sensory sensitivities who want to feel stylish — not just functional.

Needs
  • ·Clothing that works with their mobility aid or body type
  • ·Adaptive features that are invisible — not clinical
  • ·Affordable options (bespoke adaptive clothing was $300–500+)
  • ·To feel confident, stylish, and seen by fashion
Secondary User

The Artisan & Small-Scale Maker

Independent tailors, fashion makers, and small studios who want to serve the adaptive market but lack the pattern expertise and tools to do so affordably.

Needs
  • ·Accessible adaptive pattern templates
  • ·Reduced design iteration time
  • ·A new, underserved revenue stream
  • ·AI tools that don't require deep technical expertise
Top insights
  • Users don't want "medical" — they want to look like themselves, only accommodated.
  • Price, not lack of desire, is the top blocker to adopting adaptive fashion.
  • Independent makers want to serve this market but lack pattern expertise and tools.
04
Prioritization

Cost and style first — because they unlock everything else.

From 10+ user interviews and surveys through Cornell's SDS program, pain points were mapped across urgency, frequency, and impact.

Pain pointUrgencyFrequencyImpactChosen
High cost of adaptive clothingHighHighHigh
No stylish options, only clinicalMedHighHigh
Poor fit for mobility aidsHighHighMed
Lack of awareness / discoveryMedMedMed
Product decision

Solving affordability and attractive design creates the greatest immediate value for users — everything else compounds from there.

05
Solution

Three products. One inclusive supply chain.

Solution 01

AI Outfit Generator

User inputs their mobility needs, body shape, style preferences, and favorite colors. AI generates a complete adaptive outfit design with hidden accessibility features built in — and an image prompt to visualize it.

Solves: affordability + personalization + style.
Solution 02

VLM-Powered Pattern Generation

Using Vision Language Models and Python analytics pipelines to auto-generate adaptive sewing patterns — reducing the cost and expertise barrier for independent makers and artisans.

Solves: artisan cost + pattern expertise gap.
Solution 03

Curated Gallery of Adaptive Designs

A browsable gallery of AI-generated adaptive looks — low-friction discovery for users who want inspiration before committing to a custom design session.

Solves: discovery + inspiration for new users.
06
Trade-offs & Decisions

What I picked — and honestly, what it cost.

Solution matrix
SolutionBenefitLimitationComplexity
AI Outfit GeneratorScalable, deeply personalized, immediate valueRequires AI API cost managementMedium
VLM Pattern GenerationReduces artisan cost by 40%, enables supply chainComplex; requires Python pipelineHigh
Curated GalleryLow friction, fast to build, high engagementLess personalized; passive experienceLow
User-facing

Generator + Gallery

Shipped together for the direct user experience — highest immediate value, lowest friction to try.

Supply-side

VLM Pattern Generation

Powers the artisan side quietly in the backend — expands the market without adding user complexity.

07
Product Process

How I moved from a hunch to a shipped product.

  1. Step 01
    Research
    SDS + brand teardown
  2. Step 02
    User insights
    10+ interviews synthesized
  3. Step 03
    Prioritization
    Cost + style ranked highest
  4. Step 04
    Product reqs
    PRD + success metrics
  5. Step 05
    Prototype
    AI generator + gallery
  6. Step 06
    Feedback
    SDS student sessions
  7. Step 07
    Iteration
    Copy, prompts, UX polish
  8. Step 08
    Impact
    Live app · 100+ served
08
Impact

The verified numbers — and what they represent.

100+
Students Supported
40%
Cost Reduction
28–35%
Artisan Income Growth
30+
Brand Studies Conducted
Why it matters

"AdaptAI is not a fashion app. It's a statement that disability is not a design constraint — it's a design opportunity. Every person deserves to feel beautiful, confident, and seen. AI, used with empathy and intention, can finally make that possible at scale."

09
Key Learnings

What AdaptAI taught me as a PM.

Learning 01

Accessibility is a product opportunity

Over 1B people are underserved. Treating accessibility as a first-class requirement — not a feature flag — opens entirely new markets.

Learning 02

Inclusive design begins with research

Assumptions about what disabled users 'need' were consistently wrong. Only direct interviews at Cornell SDS surfaced the real hierarchy of pain.

Learning 03

AI can serve both sides of the market

The same VLM pipeline that personalizes an outfit for a user can lower artisan costs by 40%. Good AI PM means finding these two-sided leverage points.

Learning 04

What I'd test next

A closed pilot with 2–3 Cornell artisans using the VLM pattern tool end-to-end, measuring true per-garment cost, turnaround time, and repurchase intent.