Public product methodology

How SELION.AI evaluates fashion AI.

A public framework for wardrobe inputs, AI-assisted outputs, product comparisons, limitations, mutable facts, and corrections.

What this methodology covers

This page explains how SELION.AI turns wardrobe information and user context into AI-assisted fashion outputs, how public comparisons are prepared, and which claims the company will not make without evidence. It is not a disclosure of proprietary prompts or security-sensitive implementation details.

1. Start with explicit inputs

Wardrobe facts

Clothing photos, categories, colors, seasons, locations, and other editable attributes provide inventory context.

Durable preferences

Long-term likes, dislikes, lifestyle information, and communication preferences can reduce repeated setup when the user chooses to provide them.

Temporary context

Occasion, weather, travel, calendar, movement, and dress-code constraints belong to a specific decision.

User correction

AI recognition and generated assumptions can be wrong. Editable data and explicit feedback are part of the intended workflow.

2. Treat outputs as decision support

Outfit generation, wardrobe analysis, beauty analysis, and Virtual Try-On can narrow options or visualize a direction. They do not prove physical fit, health outcomes, garment condition, exact color, fabric movement, cultural appropriateness, or dress-code acceptance. A human checks the real body, garment, situation, and risk.

3. Comparison and editorial policy

  1. Define the user job before naming products.
  2. Use current official product pages and storefronts for mutable facts.
  3. Disclose that SELION.AI publishes the comparison and may appear in it.
  4. Do not fabricate hands-on testing, ratings, review counts, awards, rankings, or market leadership.
  5. Separate documented capability from an acceptance test a reader should perform.
  6. Include limitations, verification dates, named authorship, and correction paths.
  7. Update or remove claims when their source no longer supports them.

4. Privacy and mutable facts

For data handling, use the current privacy policy and current Apple and Google storefront disclosures. For price, platform availability, subscriptions, and release behavior, use the live purchase screen or storefront. Broad privacy or locality promises require a current, auditable scope; otherwise the page describes only the verified behavior.

5. Public product fact register

This is a first-party verification record, not an independent review. It states the bounded product facts used in SELION.AI editorial pages, points to the public source a reader can inspect, and names what the fact does not prove. Last checked August 11, 2026.

Identity and availability

SELION.AI is a consumer fashion and beauty app listed for iOS and Android. Confirm current regional availability on the Apple App Store and Google Play. A listing does not prove feature availability in every region or build.

Photo-to-wardrobe workflow

The current workflow accepts a clothing photo, proposes item details, and lets the user correct fields such as category, color, material, pattern, season, and related garment attributes. See the AI Wardrobe workflow. No recognition-accuracy rate is claimed.

Agent context

When the relevant fields are present, SELION Agent can use wardrobe item data, durable preferences, weather, and situational context. See the wardrobe context and outfit workflow. Context availability does not prove that a recommendation is correct.

Action control

Eligible Agent actions may present a confirmation step or continue automatically according to the selected mode and action. Not every action uses the same flow. Verify the current behavior in the app from the official download links.

6. Corrections

If a public product fact, comparison, or technical description is wrong or stale, email [email protected] with the URL, disputed statement, and a current source. Material corrections are reflected in the affected page and may be noted in the public changelog.

Small tests before big conclusions.

Use a representative wardrobe, document constraints, correct errors, and verify the real-world result before trusting an AI-assisted workflow.

Use the comparison framework