Case study · AI-assisted jewelry co-creation

EchoMade

You bring the story. AI turns it into a brief a jeweler can make.

SCROLL
EchoMade on iPhone and the jeweler's Design Workshop on a laptop
Timeline

09.2025 – 11.2025
2 months

Role

Individual project
(academic)

Tools

Figma · FigJam · Midjourney · AI prompt engineering

Surfaces

Customer mobile app
Jeweler desktop workspace

Overview

Jewelry that means something is hard to describe.

EchoMade helps people turn a personal story into custom jewelry. AI turns the story into a clear brief, and an independent jeweler designs and makes the piece.

1 · Problem

People know what a piece should mean, not how to describe it. Jewelers get feelings instead of briefs: 4–6 hours per first consult and 3–5 rounds of revisions.

2 · Research

Surveys with 40+ customers and jewelers and 12 interviews showed that both sides want to create together, with AI helping rather than taking over.

3 · Solution

A phone app that turns a story into AI concepts and a brief, and a desktop workspace where the jeweler quotes, refines and makes it. Both follow one shared order.

01 — THE PROBLEM

Feelings in, revisions out.

A custom piece starts with a story, and the story gets lost on its way to the person who makes it. Both sides describe the same gap from opposite ends.

For the customer

“I know what I want the jewelry to represent, but I don’t know how to explain it to a designer.”

For the jeweler

“Clients bring me feelings, but I need clearer starting points to turn them into real designs.”

02 — RESEARCH

Four ways of listening.

I started with the market, then asked both sides directly: surveys, interviews, and a look at what people use today.

Market research · people buy jewelry for what it means

The global jewelry market has shown consistent growth after the 2020 downturn, reaching $348B in 2023 and projected to exceed $360B by 2027.

Global jewelry market
USD billions · 2018–2027, including projections to 2027
Emotional connection is the most common reason people buy
Hover a slice
    Personalization shapes behavior
    Most companies now use AI-driven personalization; many consumers are willing to share data for tailored experiences.

    Survey · 20+ customers, 20+ jewelry professionals

    Two sets of questions, one for each side of the co-creation.

    For customers
    How comfortable are you with using AI tools to help visualize your jewelry ideas?
    What motivates you to consider custom jewelry over ready-made options?
    What challenges have you faced when trying to purchase custom jewelry?
    For jewelers
    What are the biggest challenges you face when working with clients on custom jewelry projects?
    How much time do you spend translating client ideas into initial design concepts?
    What tools or technologies would help you work more efficiently with custom jewelry clients?
    0

    wanted more control over customization than brands offer

    0

    abandoned a custom jewelry project due to complexity

    #1

    pain point for jewelers: unclear requirements

    User interviews · 6 buyers, 6 independent jewelers

    I know exactly what I want in my head, but I can’t draw. Trying to explain a specific ‘vibe’ with just words is actually super frustrating.
    Michelle24 · Student · US
    The back-and-forth emails kill me. Last time, we went through 5 rounds of edits just because they didn’t ‘get’ it the first time.
    Alvin31 · Investor · US
    I spend 6+ hours just on the initial consult trying to understand the client. If they could just show me what they want, I could focus on crafting it.
    Nihan26 · Designer · US
    0

    of buyers struggled to articulate their design vision

    4–6h

    average jeweler time on the initial consultation

    3–5

    revision cycles per project from communication breakdowns

    Competitive landscape · fast, personal, or polished — pick one

    AI design platforms
    Thenewblack.ai
    Strength
    • Fast generation
    • Low barrier to entry
    • Visual variety
    Weakness
    • Generic outputs
    • No jeweler collaboration
    • Lacks context
    Marketplaces
    Etsy
    Strength
    • Personal touch
    • Direct jeweler contact
    • Unique pieces
    Weakness
    • Time-consuming
    • Inconsistent quality
    • No guided process
    Traditional brands
    Tiffany & Co.
    Strength
    • Brand trust
    • Quality assurance
    • Polished UX
    Weakness
    • Limited personalization
    • High cost
    • Template-based
    Speed & efficiency →
    Customization depth →
    Tiffany
    & Co.Deep, but slow and costly
    EtsyPersonal, but uneven
    Thenew
    blackFast, but generic
    EchoMadeGuided co-creation
    Market gap identified

    Current solutions are either fast but generic, human but inefficient, or polished but shallow. There’s a clear opportunity for a guided co-creation journey that combines emotional storytelling, AI concept generation, jeweler refinement, and transparent pricing.

    03 — PEOPLE

    Both sides of the co-creation.

    The research pointed to two people: Aisha, who has a story, and Robert, who makes the piece. Their journeys show where each one gets stuck today.

    Journey · emotions by stage

    Tap a stage to see what they’re doing, thinking, and where the opportunity is.

    04 — KEY INSIGHTS

    AI as the translator, not the maker.

    Four things came out of the research. Together they turned the problem into one design question.

    i.
    AI as translation layer

    AI bridges communication gaps by turning emotional narratives into visual design concepts jewelers can refine.

    ii.
    Co-creation over automation

    Users want guidance and collaboration, not full automation. The goal is AI-assisted human craftsmanship.

    iii.
    Pricing transparency

    Early cost estimates tied to design complexity increase confidence and reduce drop-off.

    iv.
    Structured yet flexible

    Jewelers need structured briefs for efficiency, while customers need freedom to personalize.

    How might we

    create a platform that translates consumers’ emotional stories into actionable design briefs for jewelers — while preserving human craftsmanship and maintaining transparency around feasibility, iteration, and pricing?

    05 — THE IDEA

    From a story to a brief.

    The customer answers a few guided questions instead of facing a blank prompt. AI turns the answers into concepts and a written brief. The jeweler receives that brief, not a feeling.

    STEP 1
    Share the story
    “What’s your story?” — the flow starts from meaning, not from a spec sheet.

    4×
    STEP 2
    Guided choices
    Occasion
    EngagementWeddingAnniversaryBirthday
    Style
    MinimalVintageElegant
    Material
    GoldSilverResin
    Sketch
    Draw it together
    STEP 3
    AI concepts

    Regenerate or approve, preview, then try it on in AR.

    BRIEF
    STEP 4
    A brief, not a feeling
    Quote
    Design a futuristic, bio-morphic necklace with a continuous wavy form… translucent, amber-colored material that mimics the flow of liquid.
    Budget
    $480
    Material
    Translucent resin & gold alloy
    Style
    Futuristic & organic
    STEP 5
    The jeweler refines
    The jeweler starts in the Design Workshop with the brief beside the model, then refines, prototypes, and makes it.

    Today
    A loop of emails
    Customera feeling Jewelerguessing “it’s more of a vibe…” “is this what you meant?” 3–5×

    4–6 hours on the first consultation, then 3–5 revision cycles per project (interviews).

    With EchoMade
    One guided line
    Story Guidedchoices AI concepts+ AR Brief +early price Jewelerrefines shared visual reference from the first step

    The jeweler starts from concepts and a structured brief instead of a paragraph of feelings.

    06 — HOW IT WORKS

    Ten stages, two apps, one piece.

    Every order moves through the same ten stages on both apps. Each stage has one owner, so both sides always know whose turn it is, and money only moves once there is something concrete to agree on.

    Gate 1 · Stage 4 → 6
    Request first, pay later

    Sending a request is free. The deposit is only paid once the jeweler has read the brief and confirmed the price.

    Gate 2 · Stage 7 → 8
    Two revisions, visible

    Design approval happens on the model itself, with pinned notes. Two rounds are included and the counter is always visible, which targets the 3–5 revision cycles found in interviews.

    Gate 3 · Stage 10
    Delivery closes the loop

    Confirming delivery releases the balance and asks for a review, which then appears on the jeweler’s profile for the next customer.

    Three systems behind it · one shared platform

    The customer app and the jeweler workspace meet in a platform layer that holds the AI translation, marketplace, payments and messages.

    In the studio · from brief to a piece in your hands

    STEP 1AI processing
    STEP 2Artist refinement
    STEP 3Physical prototype
    STEP 4Metal casting
    STEP 5Quality check
    STEP 6Transit
    STEP 7Delivery
    07 — DESIGN

    A guided path instead of a blank prompt.

    Wireframes first, then a visual language, then final screens for both apps.

    Wireframes · every flow, screen by screen

    Scroll sideways through each flow. Tap any screen to enlarge it.

    Function 1 · Inspiration & onboarding
    Function 2 · AI-assisted creation
    Function 3 · Collaboration & ownership
    The jeweler’s desktop
    Dashboard
    Chat
    Project management
    Design space
    Earning analytics
    Profile page

    Visual language · liquid, glassy, a little iridescent

    Colors

    #EADDFFPrimary visual

    #6750A4Secondary visual

    #4A4458Primary text

    #79747ESecondary text

    Logo & 3D language
    EchoMade logo
    Typography
    Heading 1Quantico · 60 · Bold
    Heading 2Quantico · 45 · Bold
    SubtitlePoppins · 36 · SemiBold
    BodyPoppins · 30 · Regular
    LabelSource Sans Pro · 20 · Regular

    Final screens · from a story to a piece you can wear

    On mobile, the customer explores, shares a story, picks materials, reviews AI concepts, tries the piece on in AR, and chooses a jeweler.

    01Explore
    02What’s your story?
    03Material
    04Your EchoMade creation
    05AR preview
    06Select your jeweler
    07My creations
    08Community feed

    EchoMade desktop screens
    08 — PROTOTYPE

    The whole cycle, working.

    I built the cycle as a working prototype: the phone app and the jeweler’s workspace share one set of data, so every action on one side shows up on the other.

    Writing a story and getting four AI concepts on the phone
    01Tell the story, get four concepts and a brief
    Dragging and scaling a concept in the AR try-on
    02Try it on, pick a jeweler, send the request
    The request arrives on the jeweler desktop, a quote goes back and the customer pays the deposit
    03The request lands in Robert’s workspace. He sends a quote and Aisha commits with a deposit.
    The jeweler pins notes on the model and the customer approves the design
    04Robert pins notes on the model and sends it for approval. Aisha approves it on her phone.
    Production steps advance and the customer confirms delivery and leaves a review
    05Each production step updates the phone. Delivery releases the balance, and the review lands on his profile.
    09 — REFLECTION

    What stays human.

    EchoMade treats jewelry design as a collaboration between the user, the AI, and the maker. The AI’s job is narrow: turn a feeling into something a jeweler can hold and argue with. Everything after that stays human.

    Lesson 1
    Give AI a narrow job

    People asked for help, not automation. So the AI only writes the brief and sketches concepts, and the jeweler stays the designer.

    Lesson 2
    Always show whose turn it is

    Interviews kept coming back to email ping-pong. Giving every stage one owner makes the next step obvious on both sides.

    Lesson 3
    Price before commitment

    Two in three had abandoned a custom piece over complexity. An early estimate, then a real quote before any payment, lowers that barrier.

    Copyright © 2026 Anqi Wang