Building AI Visibility Infrastructure for Jewelry Brands — Red Pin Geek

Course Capstone · February 2026

Building AI Visibility
Infrastructure
for Jewelry Brands

in the Agentic Economy

Red Pin Geek × Andrea Li Designs redpingeek.com · andreali.com
2
AI audit agents built
8
deliverable files per client
14min
average runtime per engagement
$0
ad spend to run any of it

01 — The Problem

What We Set Out to Solve

Most OOAK jewelry brands are invisible to AI shopping agents — and don't know why.

1
AI agents can't find your products
No machine-readable specs. AI shopping agents parse structured schema fields — not description prose. OOAK jewelry is invisible to agentic discovery even when it's exactly what someone is searching for.
2
LLM crawlers are blocked by default
Squarespace blocks GPTBot, ClaudeBot, CCBot, and other AI crawlers out of the box via a single checkbox. Most jewelry brands on Squarespace have no idea this is happening.
3
No protocol feeds exist
Google's UCP and OpenAI's ACP require structured product feeds for agentic checkout enrollment. Squarespace has no native integration for either protocol. No workaround has existed — until now.
The result: OOAK jewelry brands are invisible to AI shopping agents — even when their products are exactly what a buyer is searching for.

02 — What We Built

4 components. 2 agents. 1 complete delivery system.

Agent 01

AI Visibility Audit Agent

6-category automated audit running in Claude Code. Produces Tier 1 Snapshot + Tier 2 Full Report as branded PDFs. Detects LLM crawler access, schema gaps, content architecture issues, and Agentic Commerce Readiness across any brand's website.

Agent 02

Agentic Commerce Implementation Agent

White-glove delivery agent. Input: 3 fields. Output: 8 files including schema kit, product data report, ACP feed, UCP feed, supplemental feed, and feed guide. Moves a client from NOT READY → EARLY on the Agentic Commerce Readiness scale.

Engine

Auto-Configuration Engine

Detects platform (Squarespace, Shopify, WooCommerce, Wix) from page source, generates branded SKU system (ALD-GL-BRC-001 format), discovers all collections from sitemap, and outputs a confirmation block — automatically, before touching a single product page.

Strategy

OOAK Technical Moat

Schema architecture that signals uniqueness to AI agents: LimitedAvailability + offerCount:1 + isSimilarTo (internal only) + isRelatedTo (art/collectibles). Prevents agents from surfacing alternatives.

03 — Product Suite

The Full Funnel

Each tier triggers the next. The audit finds gaps. Implementation fixes them. The re-audit proves it.

Tier 1 Snapshot
$97
Score card only
5-min read
Lead magnet & entry offer
Triggers Tier 2 upsell
Tier 2 Full Report
$597
All 6 audit categories
Complete gap analysis
Branded PDF deliverable
Triggers Implementation upsell
Agentic Commerce Add-On
$297–497
8-file implementation package
Schema + feeds delivered
NOT READY → EARLY
Triggers Re-Audit in 60–90 days
Re-Audit 60–90 Days
$597
Measures readiness progress
Built-in return path
EARLY → PARTIAL
Compounding authority value
Total funnel value per client: $97 → $597 → up to $497 → $597 — built so each engagement funds the next conversation.

04 — Under the Hood

3 Inputs → 8 Outputs → 14 Minutes

The Agentic Commerce Implementation Agent, end to end.

You Provide
WEBSITE: https://client.com
BRAND_NAME: Client Name
CURRENCY: USD
Agent Does
Detects platform from source
Generates branded SKU system
Discovers all collections
Fetches all product pages
Builds schema + feed files
8 Files Delivered
schema-kit.md
product-data-report.md + .pdf
acp-feed.csv
ucp-feed.csv
ucp-supplemental.csv
feed-guide.md + .pdf
⏱ Average runtime: ~14 minutes per client.   Readiness movement: NOT READY → EARLY

05 — Live Test

andreali.com: Before & After

AI Visibility Audit + Agentic Commerce Implementation run on andreali.com

BEFORE — Audit Findings (5 Criticals)
CRITICALLLM crawlers blocked — GPTBot, ClaudeBot, CCBot, anthropic-ai all blocked via robots.txt
CRITICALSignature Collection page missing CollectionPage schema entirely
CRITICALProduct pages missing H1 tags
CRITICALNo GTIN or MPN identifiers on any product
CRITICALNo product feed infrastructure — no native ACP/UCP integration on Squarespace
AFTER — Implementation Package Delivered
DONE6 schema blocks: Organization, CollectionPage, Product template, Product example, FAQPage, Service
DONEBranded SKU system assigned: ALD-TAM-2026-001 format for all collections
DONEACP feed + UCP feed + supplemental CSV for 5 starter products
DONEFeed guide: setup instructions for Google Merchant Center + maintenance schedule
DONEProduct data report with per-product gap checklists for all 9 products audited
Agentic Commerce Readiness:  NOT READY  →  EARLY  — one engagement, two stages of progress

06 — The OOAK Moat

Weaponizing Uniqueness
in the Agentic Economy

Schema architecture designed to tell AI agents: no alternatives exist. This piece is irreplaceable.

availability: LimitedAvailability

Not InStock — Limited

More accurate for OOAK. Signals to AI agents this is a scarce item — not a commodity with unlimited supply. Commodity schema works against you when you're one-of-a-kind.

offerCount: 1

Exactly One Unit

Declares precisely one unit exists. When an agent sees offerCount:1 + LimitedAvailability together, it reads: if the buyer waits, it's gone — forever.

isSimilarTo: [internal URLs only]

Internal Ecosystem Lock

Link only to your own archived pieces — never to competitors or external sites. Keeps AI agents inside your brand ecosystem and prevents them from surfacing alternatives in response.

isRelatedTo: [art / collectibles context]

Collectibles Repositioning

Link to fine art or rare gemstone context on your own site — not other jewelry. Repositions pieces away from commodity comparison shopping and toward art and collectibles in AI agent ranking context.

07 — Roadmap

From Invisible to Transactable

Most Squarespace jewelry brands start at NOT READY. One engagement moves them two stages.

Stage 1
NOT READY
Schema missing
No identifiers
No feeds
Stage 2
EARLY
Schema deployed
Branded SKUs
Starter feeds live
Stage 3
PARTIAL
Full catalog feed
SameAs links
GBP claimed
Stage 4
READY
Full protocol enrollment
Agentic checkout live
Platform-dependent
What moves Stage 1 → 2AI Visibility Audit ($97–$597) + Agentic Commerce Implementation ($297–$497)
What moves Stage 2 → 3Phase 4 Authority Signals: Google Business Profile, sameAs schema, directory listings
What moves Stage 3 → 4Protocol enrollment — requires platform-level support (Shopify has it; Squarespace doesn't yet)
Built-in return pathRe-Audit at 60–90 days proves progress, justifies investment, and restarts the funnel at Stage 2

08 — Key Learnings

What Building This Actually Taught Us

LESSON 01

Agents improvise — so lock down the output spec

The first test run produced files we hadn't designed — great output, but non-deterministic. We reviewed every deliverable, then hardcoded the exact structure and file names into the SKILL.md. Three versions in one session (v1 → v2 → v3) until the output was predictable every time.

LESSON 02

Auto-config removes the human bottleneck

The original agent required a manual configuration block with MPN prefix, collection codes, and platform. We reduced input to 3 fields. The agent now detects platform from source code and generates the full branded SKU system automatically — before touching a single product.

LESSON 03

A Claude Code rewrite can be an upgrade in disguise

Claude Code plan mode accidentally rewrote the SEO audit during a pivot discussion — creating something better than the original. We recovered the original, kept the rewrite, and ended up with two distinct audit products with different scope, positioning, and price points.

LESSON 04

Schema is not just for Google anymore

offerCount:1 + LimitedAvailability + isRelatedTo repositions OOAK jewelry in AI agent context — away from commodity comparison shopping and toward art and collectibles. Schema is now a competitive moat, not just an SEO checkbox.

One designer. Two AI audit agents. Eight deliverable files per client. Fourteen minutes.
Red Pin Geek — redpingeek.com Andrea Li Designs — andreali.com Built with Claude Code + Claude AI · February 2026