The Future of P2P Car Commerce
Reimagining the vehicle as a dynamic asset.
Role
Design Systems Architect & Senior Product Designer
Tools
Figma, Claude, Claude Code prototyping
Focus
Inventory Liquidity & Intent-Based UX
Format
5-day design challenge
The Supply Unlock
Traditional marketplaces run on a commitment bottleneck. Sellers are forced to list publicly and manage physical inspections before they even know their vehicle's true value. I reimagined the Kavak experience as an intent-based wealth dashboard that treats a vehicle as a dynamic asset rather than a static listing, designed to decouple valuation from commitment and turn passive owners into active marketplace liquidity.
Dashboard cards: estimated value with trend graph and AI scan CTA, buyer interest, model sell-through speed, and trade-up offers.
The Commitment Bottleneck
In the traditional P2P marketplace, the cost of entry isn't just financial, it's psychological. I identified the passive seller as the critical persona for unlocking market supply, yet this group faced the highest barriers to entry.
For these users, the legacy model imposed a "Friction Tax": the logistical burden of manual inspections and the "Inquiry Fatigue" of managing unverified strangers. By forcing a public listing before value was confirmed, the system exhausted the seller before they reached a transaction. To solve this, I architected a flow that decoupled valuation from commitment, moving from a high-friction manual process to a low-friction, automated verification system.
A seller using the AI scanner in their home garage.
The Intent Spectrum
Traditional marketplaces treat every visitor as a buyer ready to transact. This "one-size-fits-all" approach alienates the majority of the market. I designed a progressive intent spectrum to pivot the platform from a transactional funnel into a high-frequency utility.
The Intent Dashboard is the interface for that spectrum, a central hub where sellers passively track real-time buyer demand and equity, moving between Tracking, Open to Offers, and Ready to Sell without the pressure of a public listing. It's designed to be the single source of truth for sellers, consolidating valuation, demand, and transaction readiness into one coherent screen.
The Intent Dashboard on mobile, showing market intent selector, and real-time buyer matches.
The AI Camera Coach mid-scan, with a fallback option to schedule a physical inspection if the scan can't verify the car.
The AI Camera Coach
To solve the physical inspection bottleneck, I designed the AI Camera Coach: a 60-second computer vision scan intended to replace the high-friction "human mechanic" model with guided digital verification. The flow includes a fallback to a traditional physical inspection when the scan can't confirm a car's condition, so the experience stays trustworthy rather than assuming the AI gets it right every time.
The Five-Day Process
Kavak was a five-day design exercise, so how I spent those days mattered as much as what I designed. I used Claude and Claude Code across the project in different ways, not as a single "AI-assisted" blanket claim but as distinct decisions about what to hand off and what to keep.
01
Delegation
I used Claude to scan how other marketplace and fintech apps were integrating AI, comparing patterns across categories before committing to a direction.
03
Discernment
Claude worked more like a sounding board, used to pressure-test personas and the rationale for focusing on sellers over buyers, since buyer features have little utility without inventory to match against. That reasoning shaped a five-day scope built around the seller side first.
04
Diligence
The AI Camera Coach depends on real variables like lighting and connectivity, so I fact-checked its feasibility and built in a fallback to physical inspection. Claude Code stress-tested the UX flows and built the prototype I presented to Kavak's design director, along with an honest walkthrough of its time and scope limitations.
Systemic Impact
Up next
