Pattern Name
Screener

Why this exists

Pattern and collection naming is a creative process — designers draw from culture, place, and inspiration, and that's exactly as it should be. The challenge is that some names carry trademark or IP exposure that isn't obvious at the time of naming, and catching that late in the process is significantly more disruptive than catching it early.

What we're trying to prevent

Two examples from our own experience illustrate the kind of conflicts this tool is designed to catch:

Neither of these was careless — they were simply cases where there was no built-in mechanism to catch the conflict before it became a problem. This tool is that mechanism.

How it works

The screener accepts either a single pattern name or a full collection list. It runs two layers of analysis and returns a plain-English risk assessment for each name.

Layer 1 — AI Knowledge Screening (live in demo below)

The tool uses Claude (Anthropic's AI) to screen names against its training knowledge of trademarks, brand names, celebrity IP, entertainment properties, luxury fashion houses, retailer proprietary brands, and legally protected geographic designations. This catches the well-known conflicts immediately.

Layer 2 — USPTO Database Lookup (Production version)

The production version will query the USPTO trademark database in real time, surfacing active registrations, pending applications, and relevant goods/services classifications. Results are translated into plain English — no serial numbers or Class codes surfaced to end users.

Risk ratings

Design principles

• —Plain English output only No trademark jargon, Class codes, or serial numbers visible to end users. The tool translates everything into clear, direct language.
• —No explicit recommendations The tool presents facts and context. It does not tell users what to do. The risk rating communicates urgency; the action is the user's call.
• —Alternatives on every flag Red and yellow results include 2–3 alternative name suggestions that capture a similar feel without the conflict — incentivizing use of the tool early rather than after the fact.
• —Details available on request Each result has an expandable detail view and, for confirmed USPTO hits, a direct link to the live filing for anyone who wants to verify.
• —Not a legal opinion The tool is a first-pass filter, not a substitute for legal counsel. A disclaimer accompanies every result.

What the full version needs

• 01USPTO API integration Via a third-party wrapper (e.g. MarkerAPI) or a small hosted proxy — required before company-wide rollout. The direct USPTO API has browser restrictions that prevent use in a standalone file.
• 02Hosted infrastructure A simple hosted version accessible via URL — no downloads, no setup. Can be embedded in SharePoint. Requires a small API budget (Anthropic) and a lightweight server or proxy for USPTO calls.
• 03API key management In the hosted version, API credentials are stored server-side. End users authenticate via their existing Microsoft / SharePoint credentials — no API keys visible or required.
• 04Class 24 context awareness USPTO results in the textiles and fabrics class (Class 24) should be flagged with elevated urgency, as this is PK's direct competitive space.
• 05Bulk upload support Full collection lists (up to 20 names) can be pasted and screened in a single pass — returning a summary view with per-name results. Already functional in the prototype below.

Phased approach

Try it now

The prototype below is fully functional. Enter your Anthropic API key to activate, then screen any name or paste a full collection list. Try Hershey, Hampton Bays, or Versailles to see flagged results — or Riverbend or Stonehaven for clean ones.