wearepatchworks.com — AI Site Grade

Analysis

I have enough data now. Let me compile the audit.

---

The domain wearepatchworks.com is a ghost — it immediately 301-redirects to patchworks.io, yet the robots.txt lives on the ghost domain and the llms.txt 404s there, creating a fragmented AI-crawler surface.

Crawler Access

All major AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended, ChatGPT-User, anthropic-ai — receive a 200 with full HTML content on both wearepatchworks.com and patchworks.io. No UA-based blocking exists. Cloudflare sits in front with Kinsta cache. The robots.txt at wearepatchworks.com/robots.txt is a bare Yoast-generated file with a wildcard Disallow: (allow all) and zero AI-bot-specific rules. The llms.txt returns a 404 — no AI content map exists. The sitemap at patchworks.io/sitemap_index.xml is well-structured with 7 sub-sitemaps covering posts, pages, articles, case studies, features, integrations, and developer tutorials.

Content & Schema Posture

The homepage carries no Organization or SoftwareApplication schema — only WebPage, BreadcrumbList, and WebSite types. No FAQPage, Product, or HowTo markup exists despite the site having FAQ sections and pricing comparison tables. The pricing page (/pricing/) has a genuine FAQ section and a comparison table with plan features, but these are rendered as plain HTML without structured data. The /what-is-patchworks/ page, the blog, and the case-studies page return zero visible text via plain GET — content is loaded client-side via JavaScript, meaning AI crawlers that do not execute JS see empty shells. The OpenAI integration page (/integrations/openai/) contains a factually incorrect statement: "One of their [OpenAI's] notable projects is Patchworks, a software that aims to enhance AI's understanding" — this describes Patchworks as an OpenAI project, which is false and will confuse any AI model indexing that page.

Cold-Knowledge Gap

The LLM knows Patchworks as a UK-based iPaaS founded in 2015 serving mid-market and enterprise ecommerce, with 200+ pre-built connectors and positive G2 reviews (~4.5 stars). The site itself claims 1,700+ brands, 125+ connectors, an AI Assistant, and an MCP Server for connecting AI agents like ChatGPT and Claude to commerce data. The cold model has no awareness of the MCP Server, the AI Assistant, or the agentic-commerce positioning — these are the site's most differentiated AI-era features and they are entirely absent from the model's prior. The model also does not know about the Pentland Brands case study, the Castore or Mint Velvet references, or the ISO 27001 / SOC2 compliance posture.

External Signals

The DNS TXT records reveal an anthropic-domain-verification token, confirming Patchworks has proactively verified the domain with Anthropic for Claude integration. Google Workspace handles email. No Reddit threads, press mentions, or G2 review pages surfaced in web searches — the brand has surprisingly low external citation volume for a company claiming 1,700+ customers. The Wayback Machine has no snapshot of wearepatchworks.com, suggesting the domain was never meaningfully used or was recently acquired.

Surprising Findings

The wearepatchworks.com domain serves as a redirect shell with its own robots.txt but no content — any AI crawler hitting it gets redirected to patchworks.io with a 301, which is fine for indexing but means the robots.txt on the ghost domain is the one that governs crawlers that start there. The /what-is-patchworks/ page — the single most important page for AI models trying to understand the brand — is a JS-rendered empty shell (0 words extracted). The OpenAI integration page contains a hallucination-level error stating Patchworks is an OpenAI project. The site publishes dates in 2026 (e.g., datePublished: 2026-03-16), which is either a timezone/plugin bug or indicates content is being pre-dated.

Findings

1. Robots.txt lives on ghost redirect domain Medium

   The domain wearepatchworks.com immediately 301-redirects to patchworks.io, yet its robots.txt governs crawlers that start there. The llms.txt returns a 404, creating a fragmented AI-crawler surface.

   What to change: Move the robots.txt to patchworks.io and add an llms.txt file there. Remove the robots.txt from wearepatchworks.com or make it a redirect.

2. Key pages render as empty JS shells for crawlers High

   The /what-is-patchworks/ page, blog, and case-studies page return zero visible text via plain GET. AI crawlers that do not execute JavaScript see empty content, missing the most important brand-defining page.

   What to change: Implement server-side rendering (SSR) or static prerendering for these pages so that AI crawlers receive full HTML content.

3. OpenAI integration page contains false statement about Patchworks being an OpenAI project High

   The page at /integrations/openai/ states 'One of their [OpenAI's] notable projects is Patchworks', which is factually incorrect. This will confuse AI models indexing the page and damage credibility.

   What to change: Remove or correct the erroneous statement. Ensure all third-party references are accurate.

4. Homepage lacks Organization and SoftwareApplication schema Medium

   The homepage only has WebPage, BreadcrumbList, and WebSite schema. No Organization or SoftwareApplication markup exists, which limits how AI models understand the brand and product.

   What to change: Add Organization and SoftwareApplication structured data to the homepage and relevant product pages.

5. FAQ sections on pricing page lack structured data Medium

   The pricing page has a genuine FAQ section and comparison table, but no FAQPage or Product schema. This reduces the chance of rich results in AI-generated answers.

   What to change: Add FAQPage schema to the FAQ section and Product schema to the pricing comparison table.

6. Content published with future dates (2026) Low

   Some pages contain datePublished values like 2026-03-16, which is either a timezone/plugin bug or pre-dated content. This can confuse crawlers and reduce trust.

   What to change: Audit the content management system for date configuration errors and ensure dates reflect actual publication dates.

7. Low external citation volume for a brand claiming 1,700+ customers Medium

   Web searches for Patchworks reviews, press mentions, and Reddit discussions returned zero results. The brand has surprisingly low external signals, which limits AI model awareness and trust.

   What to change: Invest in PR, case study distribution, and review generation on platforms like G2 to build external citation volume.

8. AI model has no awareness of MCP Server or AI Assistant High

   The LLM's prior knowledge of Patchworks does not include the MCP Server, AI Assistant, or agentic-commerce positioning — the site's most differentiated AI-era features. These are entirely absent from the model's knowledge.

   What to change: Ensure these pages are server-side rendered and include structured data. Publish case studies and press releases that reference these features to build external signals.

What's working

• All major AI crawlers receive full HTML content — GPTBot, ClaudeBot, PerplexityBot, and others receive 200 responses with full HTML on both domains. No UA-based blocking exists.
• Sitemap is well-structured with 7 sub-sitemaps — The sitemap at patchworks.io/sitemap_index.xml covers posts, pages, articles, case studies, features, integrations, and developer tutorials, aiding crawler discovery.
• Domain verified with Anthropic for Claude integration — DNS TXT records include an anthropic-domain-verification token, indicating proactive verification for Claude integration, which supports AI agent connectivity.
• MCP Server and AI Assistant pages exist with detailed content — Pages for MCP Server and AI Assistant contain substantial text (1500+ words each) describing AI features, which can be indexed if rendered properly.
• Cloudflare and Kinsta cache provide fast delivery — The site uses Cloudflare CDN and Kinsta cache, ensuring fast load times for crawlers and users.