The Tobi AI Memo: Shopify's Bezos Moment?
TL;DR
Shopify’s leaked AI memo sets a new baseline: every employee must use AI to work 20–40% faster each year. Stefan and Jens discuss the cultural and technical shift this implies, why MCP (Model Context Protocol) is becoming essential infrastructure, and how schema-aware design prevents chaos in AI-driven development.
The AI Baseline: From Optional to Expected
The leaked “Tobi AI memo” from Shopify sparked debate across the tech world. It set a clear expectation — AI isn’t optional; it’s required. Employees are expected to show measurable productivity gains using AI tools.
If you don’t improve 20–40% a year, you’re effectively falling behind.
Jens sees this as a glimpse into the near future: every engineer will be expected to use AI copilots and agent workflows as part of their daily toolkit.
MCP and the New Developer Workflow
The team connects this shift to the Model Context Protocol (MCP) , which allows LLMs to safely call APIs and services as tools. MCP transforms local environments into programmable platforms where AI can execute real tasks, not just suggest code.
The frontend can now build itself. The model becomes a real user of your API.
This blurs the line between developer and agent, making schema-aware APIs a foundation for safe automation.
Security: The Hidden Risk Behind AI Speed
As more teams adopt MCP, Stefan and Jens warn against rushing ahead without security. Every MCP server exposes capabilities that AI can use — and misuse.
The S in MCP stands for Security.
They explain why local-first setups and supergraph-based access are safer than “vibe-coded” servers with missing authentication. Cosmo’s router embeds authorization and analytics, giving visibility into what the AI is doing.
Introducing WunderGraph Hub: Rethinking How Teams Build APIs
WunderGraph Hub is our new collaborative platform for designing, evolving, and shipping APIs together. It’s a design-first workspace that brings schema design, mocks, and workflows into one place.
The Schema as a Safety Net
Jens emphasizes that schema checks are the critical layer between AI autonomy and operational risk. Federation already tracks breaking changes and schema usage — now it extends to AI clients.
Checks aren’t just for humans anymore. They protect your agents too.
By running schema checks before an AI executes an action, teams can avoid hallucinations and broken queries before they happen.
The Dream Query Workflow
In a live demo, Jens shows how MCP can generate router configurations, OTEL integrations, and WebSocket endpoints — all from a prompt. The model interacts with the federated schema to understand what’s possible and propose changes safely.
You don’t need to write boilerplate anymore — just describe what you need.
It’s a preview of how AI-native platforms will merge IDEs, routers, and schema registries into one intelligent workspace.
This episode was directed by Jacob Javor. Transcript lightly edited for clarity and flow.
