Building Shopigent AI GEO
After Shopigent (general-purpose MCP tools) and Shopigent Returns (AI-powered returns), I wanted to build something that addresses a completely different merchant pain point — one that's emerging right now.
The insight
Traditional SEO optimizes for Google's ranking algorithm. But the way users discover products is changing. More people are asking AI assistants "what's the best X for Y?" and getting curated answers. These AI agents don't care about backlinks or keyword density. They care about structured data, clear descriptions, and real-time access.
I realized there was a gap: Shopify merchants had no way to optimize for AI discovery. The tools that exist (SEO apps, schema generators) are built for Google's world, not for ChatGPT, Gemini, and Perplexity.
The architecture
Shopigent AI GEO is a Shopify embedded app with three layers:
1. The GEO Schema Engine. At the core is a schema generator that produces JSON-LD structured data for every product in the store. Product, Offer, Rating, Breadcrumb — the full set of schemas that AI agents look for. The schemas are stored in the database and updated when products change.
2. The AI Discovery Crawler. This is the monitoring layer. It simulates how AI agents see your store by crawling web search results and recording which products appear. The dashboard shows visibility trends, top products by AI mentions, and recommendations for improvement.
3. The MCP Storefront. A full MCP server with 10 tools that let AI agents query the store directly: search_products, get_product_details, compare_products, get_promotions, get_popular_products, get_product_recommendations, search_products_multilingual, get_category_tree, and more. The same MCP tools that power AI agents also power the dashboard's product browser.
Plus: an AI Description Optimizer that scores product descriptions for AI-friendliness and gives actionable feedback.
How it fits the trilogy
Each app in the Shopigent family solves a distinct problem through the same agent-first approach:
- Shopigent (2025): "AI agents need safe, curated access to operate your store." → Confirmation gates, audit logs, 51 tools.
- Shopigent Returns (2026): "AI agents can automate the most repetitive merchant operation." → Policy engine, auto-approve, fraud detection, auto-refund.
- Shopigent AI GEO (2026): "AI agents are the new discovery channel — optimize for them." → GEO schemas, AI crawler, MCP storefront.
The common thread: MCP protocol, Remix + Polaris frontend, Prisma + PostgreSQL, deployed on Railway. Each app is a standalone product that any merchant can use independently, but together they form a complete AI-powered commerce stack.
What I learned
Building the third app is faster. The scaffolding, the MCP implementation patterns, the Railway deployment flow — all reused from the first two apps. The docs site pattern (Express + markdown, zero-build) was copy-pasted and adapted in minutes.
GEO is a new category. Most merchants don't know they need it yet. The educational challenge is bigger than the technical one. That's why the docs site and blog posts matter — they explain why GEO matters, not just how to use it.
MCP as a product feature, not just infrastructure. The MCP Storefront is arguably the most compelling feature of the app. Letting AI agents query your store directly is a powerful concept that merchants immediately understand. "AI agents can browse my store like a customer" is a better pitch than "structured data optimization."
What's next
The app is live at ai-geo-docs-production.up.railway.app. The Shopify App Store listing is being prepared. If you run a Shopify store and want to be visible to the next generation of shoppers, this is the product I wish existed a year ago.