Starbucks Runs on a Graph. Now Agents Want In.

Think about the last time you ordered from the Starbucks app. The mobile order, the loyalty points, the whole digital experience all run on a graph. The barista across the counter is on a decades-old POS.

Sharon Gorla, Lead Engineer at Starbucks, joined in 2021 to lead the graph build-out. She set out to bridge the two worlds: extending the customer graph across every retail channel and into the store. Over the years internal priorities shifted, but the architectural foundation held and, it turned out, set the team up for the rise of AI agents.

Their realization: the same graph API capabilities ideal for frontend developers (discoverability, composability, typed contracts, selective retrieval) are exactly what AI agents need. Your API schema stops being documentation for devs. It becomes the AI execution layer.

Hear Sharon on modern retail architecture, the realities of platform evolution through changing seasons, and the line she keeps coming back to: GraphQL wasn't built for AI — but it turns out AI was built for GraphQL.

What you'll hear

• Why the same graph you built for customers is what AI agents want

• How to play the long game through the seasons of platform investment

• Why MCP in front of your graph beats hand-wiring agent tools