VALIDATE FIRST · Concept, not a live product
Wardrobe Copilot is a proposed way to retrieve a proven outfit from clothes already owned—with reasons, not invention. The independent review found the idea coherent and the evidence insufficient.
The risky assumption
“Choosing what to wear happens often enough to deserve a product.”
The reviewed truth
The repository proves a useful browsing surface and publication pipeline. It does not prove that ranked outfit help is wanted, safe to expand, or worth maintaining.
Public on Site and an image to enter the current artifactThis baseline is unauthenticated, owner-identifying, and currently includes personal photos and source identifiers. Its public posture needs a separate owner decision.
These remain proposed capabilities. This page does not simulate their availability or ask for credentials.
The job to investigate
Current workaround
The friction sounds plausible. It has not yet been observed in a real dressing workflow.
V0 · no new product
Validation is deliberately manual. A failed gate means simplify, pivot, or stop—not add technology.
Run a consented seven-day diary. Continue only with at least three genuine decision occasions.
Serve five real scenarios by hand. Continue only when four of five suggestions are wearable without correction.
Log repair effort. Continue only at an average of 15 minutes or less per useful decision.
Audit the Notion schema read-only and prove five redacted whole-look fixtures. No source changes.
Grounded product view
This local visualization separates today’s archive, the manual validation step, and a conditional future experience. It is not connected to wardrobe data.
Implemented component
Search and filter approved outfit records. The interface retrieves cards; it does not interpret a dressing context.
Research only
Use a real occasion and existing proven looks. Observe whether the answer is accepted and whether data repair erases the benefit.
Planned only if gates pass
A future private view could rank whole looks deterministically. Every result would show reasons, unknowns, and freshness—never an unexplained confidence score.
Proven look / fixture
Matches: requested occasion · recorded dress code
Unknown: current availability is not tracked
The hypothesis, not the moat
Competitors already offer wardrobe capture, suggestions, feedback, planning, sharing, and export. The only credible test is whether this narrower posture works better for the owner.
Trust is a product surface
Public read, private read, draft, source write, publish, delete, and operations are separate authority classes. A convenient label never lets one imply another.
Agent / MCP · deferred R3
If the web flow proves useful, a later read-only wardrobe.* surface may list, search, get, or recommend through typed, owner-scoped calls. Draft tools would require a new decision. Publishing and deletion remain outside the agent path.
Local validation note
This single note rehearses the first diary question. It does not join a waitlist or send anything anywhere.
Local prototype. Your response is stored only in this browser on this device. No analytics, cookies, account, or network submission.