Case Study: Shopify
Overview
| Field | Value |
|---|---|
| Company | Shopify |
| Industry | E-commerce Platform |
| Headquarters | Ottawa, Canada |
| Deployment Date | 2025 |
| MCP Version | Anthropic MCP spec |
| Enterprise Readiness Score | ⭐⭐⭐⭐ |
Use Case
Shopify's Sidekick AI assistant is powered in part by MCP tooling that connects the agent to merchant store data — products, orders, customers, analytics, and theme settings. Merchants can ask Sidekick to update product descriptions, generate discount codes, analyze sales, and manage inventory without leaving the admin interface.
Connected Systems
| System | Type | Direction | Auth Method | Notes |
|---|---|---|---|---|
| Shopify Admin API | Store Management | Read/Write | OAuth (merchant-scoped) | Products, orders, customers |
| Shopify Analytics | BI / Reporting | Read | OAuth | Sales and traffic metrics |
| Shopify Themes | Storefront | Read/Write | OAuth | Theme customization |
| Shopify Inventory | Operations | Read/Write | OAuth | Multi-location stock management |
Architecture Pattern
Each merchant's Sidekick instance is scoped to their own store data via merchant-level OAuth tokens. No cross-merchant data leakage.
Governance Controls
- Merchant-scoped tokens: Each agent session bound to a single merchant's OAuth credentials
- Scope restrictions: Sidekick requests minimum API scopes per action type
- Human confirmation: Destructive actions (bulk delete, price changes) require merchant approval
Outcomes
- Merchants report significantly reduced time to create and update product catalogs
- Sidekick handles common admin tasks that previously required developer intervention
- High adoption among non-technical merchants on Plus and Advanced plans
Source Links
Evidence Quality Notes
First-party MCP server with public repository. Sidekick is a GA product with public documentation.