Author: Tom Hipwell, VP of Engineering at Nory
Everyone is talking about AI, and the hospitality industry is no exception.
Large Language Models (LLMs) and other forms of AI, we’re told, are going to revolutionise everything from ordering and scheduling, to forecasting and delivering.
But as someone who has spent a lot of time over the past decade building and thinking about how these technologies fit into the complex, fast-paced, frenetic world of a restaurant, I’ve come to a realisation: Simply applying general-purpose AI is not enough.
The real challenge in hospitality isn’t just about generating reams of text or crunching page after page of numbers; it’s about understanding the industry context.
And what is that context? It’s one we all know – that the industry is a web of interconnected variables where a change in one area – a sporting event, for example – can send ripples through everything from revenue forecasts and inventory management, to hourly staff scheduling.
Imagine if every job in your restaurant was done by a different person, and none of them communicated with one another.
Your waiters don’t talk to the bartender. The head chef and sous chef operate in silence. The maître d’ doesn’t even look at the general manager.
It sounds absurd, but it mirrors the state of many hospitality tech stacks today. Your POS doesn’t talk to your labour management system. Your inventory platform has never met your labour management tool. You’re running six different log-ins just to understand if yesterday was profitable.
It’s not a recipe for a successful business of any kind, let alone one in hospitality.
And it’s about more than simple inefficiency; it’s a matter of survival. Rising labour costs, high energy rates, less spending on dining, and numerous other pressures mean you’re already battling for every single percentage point of your margins.
In this context, I would argue that we need to be talking about AI less as a general, all-purpose blunt object, but as a precise, specialised, finely tuned tool.
Where general-purpose AI Doesn’t Work
An LLM, the technology behind tools like ChatGPT, is a brilliant communicator. It can summarise a complex document or explain a tricky concept pretty well.
But ask it to build a precise rota for 200 restaurant locations, taking into account staff holidays, local regulations, prep-time requirements, while also optimising for efficiency metrics like sales per labour hour (SPLH), and you’ll quickly see its limitations.
General, all-purpose AI doesn’t have the granular, specialised precision that modern hospitality businesses need. For me, that’s a system with a truly precise understanding of how a restaurant functions across all its sites. It’s a system that understands how everything in a hospitality business is connected, from kitchen prep sheets, to front of house workflow, to compliance, and everything in-between.
Integrating and analysing these complex, interconnected dynamics and operational dependencies is what can unlock the power to drive margin improvements. That’s the belief that drove our thinking at Nory: The way to achieve effective specialisation lies in combining three core technologies that work together: LLMs, Machine Learning (ML) algorithms, and the specialised agents that put them into action.
What We Mean By AI
Let’s talk about the terms we use when we talk about AI. And then let’s look at how that ties into the way we can think about an AI operating system.
An algorithm is like a recipe to get something done. It’s the precision engine. The distinction with AI is in how that recipe is developed.
Instead of following a set of well-defined, predetermined steps, ML algorithms learn and refine the optimal recipe themselves, by analysing training data (like sales patterns, prep times, and compliance rules) to generate the most optimal outcome.
LLMs serve as the communicators. They are essential for explaining the complex output of the ML algorithm, allowing a manager to chat with the model and turn dense data into conversational, accessible insights.
The AI agent is the functional layer, the executor. An agent takes the precise decision, acts like a dynamic supervisor to plan steps and complete more complex tasks across the entire operating system.
Let’s Talk About Agents
When we’re talking about restaurants, AI agents are the true executors. An agent isn’t just a trigger; it’s an AI that can actually understand a goal, create a plan to get there, and then successfully execute that plan.
Traditionally, automation was rigid, with every step specified: If this, then that.
But in the new world, agents navigate that operational complexity for you. That’s more than just doing the first step; that’s managing the whole process to achieve a specific outcome.
Picture it: It’s Thursday afternoon. Your general manager at one site just texted; they’re down two servers for Saturday night and there’s a big local match on. That’s an hour of phone calls, checking availability, and worrying about overtime rules.
In the new world, your forecast already shows an increased expected sales volume, automatically reviews the staff pool, considers holidays and regulations, identifies the optimal coverage from available staff, secures confirmation, and updates the live schedule. It takes seconds. And the manager is only needed to rubber stamp the result.
It’s only possible when all your data and tools are brought together in one unified environment, what we call an AI Operating System (AI OS).
The Power Of An AI Operating System
The power of this unified approach lies in continuous, cross-functional feedback loops. In an AI OS, every operational step, from sales forecasting to rota planning and stock ordering, is viewed not as a standalone item, but as part of a single, interconnected profitability loop.
Every decision affects your margin. But when your systems actually talk to one another, these problems solve themselves. The operating system delivers genuine business resilience and has an effect on your margins because every stage of the loop feeds the next one.
Better forecasts mean more efficient labour scheduling or more accurate demand-based ordering. And having a single forecast that connects all of these dots is even more powerful. You review one forecast, tweak it with your specialist on-the-ground knowledge, and everything that depends upon that model also updates its expectations.
For too long, hospitality operators have been battling against the clock, manually connecting the dots between disjointed systems. A specialised, context-aware AI OS offers the control and visibility needed to streamline operations and unlock margin improvements. It allows operators to step back from the constant operational firefighting and focus on what they do best: Serving their customers.
And it’s why, at Nory, we’ve built that single, unified operating system to work with your existing tech stack, generate those unique margin-boosting insights about your business, and automatically action them for you, so you can focus on what you do best: Serving your customers.
Ready to see what the power of an AI operating system can do for your business?
