AI Heads, LLC · Pre-launch product
SoleStamp
Turning a phone camera into a guided sneaker assessment and collection experience.
SoleStamp helps sneaker owners move from an unidentified pair to a structured record of its model, condition, AI-generated authenticity guidance, estimated market value, and place within their collection. I designed and built the product from 0 to 1, directing Claude Code as an agentic implementation partner across product strategy, UX/UI design, architecture, development, debugging, and testing.
SoleStamp is currently in pre-launch testing. Its authenticity and valuation outputs are informational AI-generated assessments—not guarantees, certifications, or professional appraisals.
- Role
- Lead Product Designer & AI Product Consultant
- Company
- AI Heads, LLC
- Product
- SoleStamp
- Platform
- iOS and Android
- Build
- Expo · React Native · TypeScript · Supabase
- Status
- Pre-launch testing
Project at a glance
- Challenge
- Sneaker owners often have to move between identification tools, condition judgments, authenticity services, resale listings, and spreadsheets to understand a single pair.
- My role
- I owned product strategy, workflow and information architecture, UX/UI design, design-system direction, AI behavior, technical direction, implementation, and testing.
- Product scope
- A guided mobile workflow connecting photo capture, identification, condition assessment, authenticity guidance, valuation, collection management, portfolio analytics, comparison, and release discovery.
- Current state
- The product is undergoing pre-launch iOS and Android testing. No post-launch product or business outcomes are available yet.
Overview
A connected assessment, not a collection of disconnected tools
The sneaker resale market has a trust problem before it has a pricing problem. Identifying a model does not tell someone whether a specific pair appears authentic, how real-world wear affects its condition, or what it may be worth in its current state. SoleStamp connects those questions in one guided experience while making the limits of AI-generated guidance visible.
Stage 01
Capture the evidence
Guide the user through the photos and information needed to assess a specific pair.
Stage 02
Build the assessment
Identify the sneaker, evaluate visible condition, and provide clearly qualified authenticity guidance.
Stage 03
Make the result useful
Connect the assessment to valuation, collection tracking, comparison, portfolio analytics, and release discovery.
Assessment flow
From camera capture to collection record
The primary workflow progressively gathers evidence and turns it into a structured sneaker record. Results are presented as guidance, with visible qualifications where the product cannot make a definitive determination.
Step 1 of 7: Start a scan
Step 1 of 7
Start a scan
Start with the camera or photo library. Light and dark themes preserve the same hierarchy and primary action.
Collection
Turning assessments into an owned collection
Collection states
Collection views provide a persistent record of assessed pairs across free, Pro, list/grid, and theme variations.
Portfolio analytics
The Pro portfolio dashboard brings together invested value, current estimated value, return calculations, and value history. This is a working subscription-gated experience, not a claim that users achieved a specific financial return.
Verified Pro behaviors are limited to: more than the free allowance of three scans per month, the Portfolio Dashboard, and the Net Worth Over Time valuation-history chart.
Compare
Compare helps users review two saved pairs side by side across value, condition, authenticity guidance, return, and date-added information. The replacement search opens in a focused bottom sheet. Compare is limited to two items for every user.
Product system
An archival ledger rather than a typical resale app
SoleStamp uses a warm-paper foundation, ink-toned hierarchy, restrained borders, minimal elevation, and editorial typography to make each record feel closer to documented provenance than a marketplace listing.
Color
- paper
- paperRaised
- ink
- inkSoft
- rule
- forest
- oxide
- brass
- amber
Typography
- Newsreader — display
- Public Sans — interface and body
- IBM Plex Mono — prices and data
Themes and profile
Light and dark themes preserve content hierarchy and action placement.
Accessibility decision
The light-mode brass text token was darkened after contrast measurement showed the earlier value did not meet WCAG AA against brass-tinted chip backgrounds. The revised value reaches the required contrast range.
The core scan-to-valuation flow still needs a systematic accessibility pass. Accessibility labels and roles are not yet consistently implemented across those screens.
Release discovery and reminders
A supporting discovery experience helps users browse upcoming sneaker releases and request a reminder for a selected launch. It extends SoleStamp beyond assessing owned pairs without interrupting the core assessment workflow.
Responsible AI and data boundaries
01
Qualified results
Authentication uses likely authentic, likely counterfeit, and needs review—not guaranteed authentic or certified.
02
Market-data hierarchy
Valuation attempts external market-data sources first and falls back to a clearly labeled AI-generated estimate when current listing information is unavailable.
03
Private image storage
Scan images are stored in a private, user-scoped Supabase storage bucket with access policies restricting users to their own folders.
04
Current deletion limitation
Deleting a collection item or account does not yet remove all stored scan photos. Complete storage-file deletion is a known pre-launch data-lifecycle requirement that is not implemented yet.
Delivery
From product strategy to testable mobile builds
- My contribution
- I owned the product strategy, workflow and information architecture, UX/UI design, design-system direction, AI behavior, technical direction, implementation review, debugging, and pre-launch testing.
- AI-assisted implementation
- I used Claude Code as an agentic engineering tool across architecture, feature implementation, code review, debugging, and verification. I directed its work, reviewed its output, challenged unsupported assumptions, and verified changes against source code and physical-device behavior.
- Evidence-first debugging
- During TestFlight testing, an iOS subscription-related launch failure required investigation beyond local checks and summarized stack traces. The process included reviewing symbolicated device evidence, checking upstream issues, and verifying behavior against compiled source. The subscription layer was moved from a third-party purchase SDK to direct StoreKit integration with server-side notification handling. A separate cold-launch crash remains under investigation on multiple physical devices.
- Current status
- SoleStamp is in pre-launch testing. The iOS build is being tested through TestFlight, and Android has been distributed internally through an EAS APK. The core workflow has been exercised on physical devices, but release readiness has not been declared and no post-launch results are available.
Reflection
Building SoleStamp reinforced that directing AI-generated implementation still requires product judgment, source verification, technical accountability, and real-device testing. A change is not complete because it appears correct in isolation; it has to survive the workflow, the platform, and the conditions in which people will actually use it.
