01 Overview
Why UX designers spend more
time on busywork than design
The average UX designer spends only about 30% of their week actually designing. The rest is consumed by synthesizing user interviews, writing meeting notes, generating throwaway variations for stakeholder buy-in, formatting research reports, and re-explaining design decisions across reviews. The most valuable skills — taste, judgment, problem-framing — get squeezed into the margins.
This case study is a research-only investigation: I set out to understand whether an AI co-pilot for designers is a real opportunity, and if so, what shape it should take. No product was built. The output is a problem framing and a set of design proposals grounded in evidence — a foundation for a future product team to act on.
"Before designing a solution, design the question. AI for designers is everywhere — but the right AI for designers is still an open problem."
— Research premise
68%
of designers report spending too much time on non-design tasks
14hrs
Weekly average spent on research synthesis & documentation
3.2×
More iterations possible when ideation is AI-assisted
02 Understanding the User
What designers actually want
from AI in their workflow
To make sure this research wasn't another solution looking for a problem, I spent four weeks listening. I ran contextual inquiries, interviewed UX practitioners across seniority levels, and looked closely at how existing AI tools (Midjourney, Galileo AI, Uizard, Magician for Figma) actually fit — or didn't — into real workflows.
🎙
Designer Interviews (11)
60-min sessions across junior, senior, and lead UX roles. Focused on workflow pain points, attitudes toward AI, where they trust automation, and what would make them adopt a new tool.
📊
Survey (62 designers)
Quantitative study on time allocation per task, current AI tool usage, frustrations, and willingness to pay. Revealed clear demand for research synthesis and ideation support — not pixel generation.
👀
Workflow Observation (7)
Shadowed designers during real project work to spot moments of friction — context switching, dead-ends, repetitive tasks. Many problems weren't articulated in interviews but visible in screen recordings.
🔍
Competitive Analysis
Evaluated Galileo AI, Uizard, Magician for Figma, Khroma, and Diagram. Identified gaps: most tools generate visuals but don't help with research, critique, or design rationale — the actual hard parts.
Critical Research Finding
Designers don't want AI that generates final designs. They want AI that accelerates the thinking around the design — research synthesis, idea expansion, critique, pattern lookup. Pixel-generation tools were the most-tried and least-retained category. The unmet need is cognitive, not visual.
User Persona
👩💻
Maya Iyer
Senior UX Designer · 32 · Bangalore
Maya leads design for a fintech product. She's juggling 3 active projects, mentoring 2 juniors, and prepping for a stakeholder review every other week. She loves the design craft but resents the hours lost to transcribing user interviews, generating throwaway variations for buy-in, and writing the same critique notes in every review. She tried Galileo AI and Uizard but found them gimmicky — they made screens she didn't want.
5+ years experience
Manages multiple projects
Skeptical of generative AI hype
Values craft, hates busywork
Wants tools that respect taste, not replace it
Maya's needs boiled down to three: cut the time spent on research synthesis, get fast ideation variations she can edit, and have a smart critique partner when she's working alone late at night.
03 Problem Definition
Translating pain points
into design questions
From the research synthesis, I identified five core problem areas in the UX design workflow and translated each into an actionable "How Might We" question — the framing a future product team could use as a starting point.
Pain points identified:
- Research synthesis overload — 8-12 hours to manually code and theme one round of user interviews
- Ideation bottleneck — designers default to first idea due to time pressure; rarely explore enough variations
- Critique vacuum — solo designers and remote teams lack fast feedback loops; mistakes ship
- Pattern recall — designers can't remember every reference they've seen; reinventing solved problems
- Documentation drag — design decisions don't get captured; rationale lost when designer moves teams
How Might We questions:
HMW turn raw interviews into themes in minutes, not days?
HMW expand design exploration without expanding effort?
HMW give designers fast, honest critique on demand?
HMW surface relevant references when designers need them?
HMW preserve design rationale without writing more docs?
"Be the second pair of eyes the designer wishes they had — not the hand that takes the pen."
— Design principle to carry into the solution
04 Proposed Direction
What a solution
could look like
Based on the research, I've put forward a directional proposal — not a product, but a set of principles and capability ideas that a future design team could build against. Each suggestion ties directly back to a pain point identified in Section 03.
PROPOSAL 01
A native plugin, not a standalone product
Designers won't leave Figma. Every observed designer kept Figma open as their primary workspace, with Slack and the browser as satellites. The right form factor is a sidebar plugin that lives inside the design tool — discoverable from the same panel where Inspect and Comments already live. No new app to learn, no login fatigue, no context switch.
PROPOSAL 02
A research synthesis engine — the highest-value use case
Drop in 12 interview transcripts, get themed insights, pain-point clustering, persona drafts, and suggested HMW questions in minutes. This is where designers most clearly said yes. It's the heaviest unpaid labor in the workflow, and the most quantifiable win. Should ship first as the wedge into the product.
PROPOSAL 03
Variations as editable starting points, not finished screens
Generated outputs should be explicitly framed as starting points — labeled, parameterized, and editable in Figma layers from the moment they're added. Each variation should explain its design rationale ("Bold typography · High-energy color") so the designer learns, not just copies. The fastest way to lose a designer's trust is to feel like you're trying to replace them.
PROPOSAL 04
Critique as transparent scoring, not opaque grading
Provide a holistic score (e.g., 82/100) with a breakdown across accessibility, visual hierarchy, and consistency. Show specific, numbered suggestions on the canvas itself — color-coded by severity. Avoid single grades that feel like judgment. The goal is to be useful, not authoritative.
PROPOSAL 05
An explicit "Apply" gate — AI never edits the canvas autonomously
This emerged as the single strongest signal from research. Designers hate AI that changes things without permission, even when correct. Every AI output should be a suggestion, and the designer must press Apply to commit it — with a short Undo window. Slower? Yes. Trusted? Massively.
PROPOSAL 06
A curated reference library, not the open web
Pattern suggestions should pull from a vetted, licensed library rather than scraping competitor products or Dribbble. This protects against quality issues, IP risk, and the "everything looks the same" failure mode of current AI design tools. Worth investing in the library as a competitive moat.
Closing Note
These proposals are directional, not prescriptive. The intent is to give a future product team a clear, evidence-backed starting point — a shared understanding of what designers want, what they don't, and what would actually earn a place in their workflow. The most-loved AI for designers won't be the most powerful one. It'll be the one that respects them most.