AI Disclosure

For each AI call, only the minimum data needed for that feature is sent:

• Outfit generation: your Style DNA fields, weather, occasion, and any anchor/closet pieces relevant to the compose. Not your email, full name, payment info, or full closet inventory beyond what's referenced.
• Stylist chat: your message + recent thread context + a compact profile brief. Each thread is stateless from the provider's perspective.
• Photo recognition: the photo you uploaded + a prompt asking for a catalogue entry. The image is processed and discarded by the provider per its enterprise zero-retention policy.
• Virtual Try-On: your body photo + a garment image. Both are sent to fal.ai for compositing, then discarded by them after the synthesis completes.

All AI providers we use are configured under enterprise API agreements that:

• Prohibit training their models on Tailr customer inputs or outputs.
• Apply zero-retention or short-retention policies (typically 30 days or less, for abuse-detection only).
• Process data in the United States; transfer mechanisms (where applicable) are governed by the Privacy Policy.

Within Tailr, we cache outfit composes and chat replies briefly (5 minutes to 24 hours, depending on the surface) to speed repeat reads and to allow you to revisit a result. Cached entries are scoped to your account and purged on the retention schedule in the Privacy Policy.

What it's good at

• Translating your Style DNA into a complete head-to-toe look.
• Adapting outfits to weather and occasion at scale.
• Surfacing brand-specific products that match a vibe you described.
• Answering general styling questions (fit, color, occasion etiquette).

What it's not

• A professional stylist, tailor, or fit consultant. The AI doesn't see you wear the clothes.
• A guarantee of in-stock or current pricing. Brand catalogs change minute-to-minute and we cache them in 5-minute to 4-hour windows.
• A formal dress-code arbiter. For high-stakes events (weddings, interviews, religious occasions), confirm against the actual dress code or organizer.
• A safety or medical authority. Don't use AI suggestions in place of allergen, weather-survival, or workplace PPE advice.

The underlying LLMs are trained on broad internet text; they can reflect biases present in that data — gender, body type, cultural style, brand visibility, and others. Tailr applies several layers of mitigation:

• A strict gender filter on brand catalogs so a man's outfit doesn't surface women-only pieces (and vice versa). Both men's and women's coverage spans 1,000+ brands.
• Server-side strippers that drop pieces with no product image, off-occasion swimwear, and weather-inappropriate outerwear.
• An ongoing accuracy bar — outfits failing occasion, weather, fabric, or proportion checks are caught + re-rolled before they reach you.
• Per-user Style Genome drift that learns from your likes/dislikes and biases future composes accordingly.

If you notice an outfit that misrepresents your taste, body, occasion, or any group: tap the feedback button in the bottom-right of any page, or email support@tailrstylist.com. We review user reports promptly.

You always remain in control. Every AI suggestion can be regenerated, edited, or ignored. There is no automated decision that has legal or similarly significant effects on you within the meaning of Article 22 GDPR.

Each outfit compose runs a single LLM inference (a few seconds of GPU time). Virtual Try-On runs a single image-diffusion inference (typically 15–30 seconds of GPU time). We cache aggressively to avoid re-computing identical requests. Aggregate compute use is published in our annual transparency report if you want details — email support@tailrstylist.com.

Found a harmful, offensive, or clearly wrong AI output? Use the in-product feedback widget (bottom-right of any page) or email support@tailrstylist.com. We treat reports of biased, unsafe, or impersonating outputs as priority issues.

We'll keep this page current as we change models, providers, or features. Material changes are flagged with a revised “Last updated” date and (for substantive shifts) an in-product banner.