The restaurant industry stands at a pivotal inflection point where AI and technology have become inseparable, according to a recent episode of the Fast Casual Nation podcast hosted by Paul Barron and Cherryh Cansler. With 79% of restaurant operators either implementing or considering AI solutions, the conversation has shifted from “if” to “how” brands can leverage this transformative technology. Joining the discussion were Kerry Leo, VP of Technology at Shipley Donuts, and restaurant tech consultant Paul Molinari, who shared real-world insights into AI implementation and its measurable impact on operations.
Shipley Donuts’ recent rollout of an AI-powered online ordering system with Saivory demonstrates the practical applications driving early adoption success. The 90-year-old brand saw impressive results out of the gate: 22% of customers used AI features, with AI-assisted orders generating 24% higher average order values. Leo explained that the system was deliberately designed to replicate the experience of Shipley’s best cashiers, incorporating suggestive selling and intelligent features like auto-boxing for donuts. When a customer selects 11 donuts, the AI prompts them to add one more for a dozen discount—a natural upsell that feels helpful rather than pushy. This customer-facing success came after Shipley had already been using AI on the backend for inventory management, prep prediction, and helping new locations in unfamiliar markets understand what to make.
The conversation revealed a critical distinction between early AI failures and today’s more sophisticated implementations. McDonald’s 2019 acquisition of voice ordering technology, for instance, was plagued by accuracy issues that made headlines for all the wrong reasons. Fast forward to today, and the technology has matured significantly. Leo shared that Shipley is now preparing to integrate voice ordering into their mobile app and exploring drive-thru applications, though he emphasized the extensive testing required for the latter. The key difference is that modern AI systems learn from data rather than simply following rigid programming, allowing them to adapt to specific locations, available inventory, and customer preferences in real-time.
Data unification emerged as the cornerstone of successful AI implementation throughout the discussion. Leo described Shipley’s Snowflake environment, where data from their website, point-of-sale systems, and transaction-level details all feed into a single source. This unified approach enables the kind of sophisticated analysis that was once impossible—like asking why a specific location’s sales differ on Tuesday at 7:31 PM compared to previous years. Molinari emphasized this concept of a “single pane of glass,” where data is ingested, cleaned through pattern recognition, and made accessible to whoever needs it. The goal is prescriptive action: alerts sent directly to store managers about what needs fixing, suggestions on schedule adjustments based on real-time sales, or recommendations to make more strawberry iced donuts because inventory is running low compared to typical demand.
The podcast also tackled the elephant in the room: why aren’t more brands moving faster on AI adoption? The answers ranged from cost concerns to the classic challenge of disparate systems that don’t communicate with each other. While enterprise brands like Domino’s can partner directly with Microsoft to reduce training time from 16 hours to under one hour and improve forecasting accuracy from 75% to 95%, smaller chains must take a more piecemeal approach with point solutions. However, as Molinari pointed out, agentic AI—where AI agents communicate directly with each other—may soon solve integration challenges without requiring complex API protocols. Leo confirmed that the technology providers Shipley works with are already discussing this future, where systems converse without traditional web hooks.
Looking ahead, the restaurant industry appears poised for a fundamental shift in how customers discover and order food. Leo predicted that traditional Google search is “done,” replaced by answer engine optimization where brands must be embedded in large language models like ChatGPT, Gemini, and Claude. Shipley’s partnership with Saivory positioned them to enter ChatGPT’s new marketplace alongside DoorDash, enabling AI-to-AI ordering in the near future. For operators still on the sidelines, the message was clear: start with specific problems you’re trying to solve, choose partners with strong APIs and integration capabilities, and recognize that AI implementation is a journey. As Leo candidly assessed, Shipley rates themselves a five out of ten on AI adoption—confidently moving forward while acknowledging there’s much more to come on their roadmap.
