Max Wu
← Back to Work

AI Decision Support · Used-Car Experience

Crossover Car

Turning used-car uncertainty into a confident shortlist.

An AI-powered decision-support platform that helps buyers move from real-life needs to an evidence-backed shortlist—while keeping the choice in their hands.

Crossover CarAI decision-support concept
A clearer way
to your next car.
01AutoMatchWhat fits my life?
02AutoCheckIs it worth the risk?
03ComparisonWhat fits my priorities?

01 · The Decision Gap

Finding a car is easy. Knowing whether it is right for you is not.

“119k miles on a 2020 car—is this a good deal or a future repair bill?”

First-time buyers can browse thousands of used-car listings in minutes. But comparing prices, interpreting vehicle history, and understanding long-term risks requires knowledge many buyers do not have. More options often create more uncertainty—not more confidence.

Designed for first-time used-car buyers with limited automotive knowledge.

02 · What Research Changed

More information was not creating more confidence.

Customer voice“I just want someone to clear the noise and tell me what actually fits my life.”
01

Buyers already understood their basic needs.

Most participants could describe their budget, commute, passengers, and lifestyle without difficulty.

02

They lacked a translator.

The challenge was connecting real-life needs with vehicle types, features, costs, and risks.

03

Confusion peaked after users found several options.

The lowest point occurred during qualification and comparison—when buyers had enough information to feel overwhelmed, but not enough understanding to move forward.

Card sorting synthesis

The information buyers needed fell into three clear groups.

Financials

  • Asking price
  • Insurance
  • Taxes and fees
  • Ownership cost

History & Trust

  • VIN and title status
  • Accident history
  • Maintenance
  • Repair records

Core Specs

  • Year and model
  • Mileage
  • Fuel economy
  • Vehicle type

From Research to Product Strategy

We stopped designing another car marketplace.

This insight shifted the concept from vehicle browsing to decision support.
How might weReduce decision overload while keeping the reasoning visible and the final choice in the buyer’s hands?
What fits my life?

AutoMatch

Three personalized recommendations

Is this car worth the risk?

AutoCheck

Score, risk signals, and plain-language explanation

Which option fits my priorities?

Comparison

Priority-based comparison and trade-off summary

MatchEvaluateComparePrepare for a test drive

03 · AutoMatch

Turn vague preferences into a useful shortlist.

Problem

Buyers often start with a budget and daily needs, but not the vocabulary to turn them into vehicle criteria.

Decision

AutoMatch structures the conversation in five stages, moving from context to constraints, priorities, and a three-vehicle shortlist.

01Context Summary
02Basic Constraints
03Lifestyle Context
04Personal Priorities
05Three-Vehicle Shortlist
01

Start with life, not specifications

Users begin with familiar lifestyle needs instead of automotive terminology.

02

Confirm before recommending

The system summarizes the user’s context so misunderstandings can be corrected early.

03

Limit and explain the shortlist

Three recommendations keep the result manageable, while each recommendation explains why it fits.

04 · AutoCheck

A score users can understand—not just accept.

Problem

A single vehicle score looks simple, but it can hide the evidence needed to understand risk and plan the next step.

Decision

AutoCheck pairs the overall result with visible evidence, severity cues, and practical actions.

The result moves from an overall judgment to supporting evidence, critical risks, and recommended actions.

AutoCheck · Interaction structure
  1. 01 / OrientOverall score & risk level
  2. 02 / UnderstandBreakdown & supporting evidence
  3. 03 / ActExplanation & next steps

05 · Comparison

Compare trade-offs, not columns.

Problem

Traditional tables show that two vehicles are different, but rarely explain which differences matter for the buyer’s priorities.

Decision

Comparison turns saved AutoCheck results into a focused workspace with adjustable priorities and a plain-language summary.

01

AutoCheck History

02

Select two vehicles

03

Adjust priorities

04

Review trade-offs

Selection and defaults
  • Begin with two vehicles from AutoCheck History
  • Add one optional third vehicle
  • Use balanced weights until priorities change
Adjustable priorities
  • Reliability
  • Price fairness
  • Mileage
  • Maintenance history
Example trade-off summaryThe Honda CR-V offers stronger reliability and maintenance history, while the Mazda CX-5 is more affordable and has lower mileage. The CR-V may be the safer long-term option; the CX-5 may better fit a tighter upfront budget.

06 · Outcome & Reflection

What the final prototype delivered.

The final prototype created a structured path from vague needs to a shortlist users could evaluate, understand, and compare.

Concept Project · Interactive High-Fidelity Prototype
Reflection

Designing trust meant leaving room for doubt.

At first, we treated AI as a shortcut to the answer. The stronger direction was to surface the evidence, explain the trade-offs, and let the buyer decide. This project showed me that explainability is not an additional feature—it is part of the core interaction.
What we would test next
  1. Validate the scoring framework with automotive experts
  2. Connect real listing and vehicle-history data
  3. Test score comprehension and decision confidence with first-time buyers

Next Project

ClipsInsight

View Next Case Study →
Back to Selected Work