The talk covered how the technology has evolved, where it sits today, and what it looks like when it's actually useful inside a hospitality business. Here’s our key takeaways.
There's a reason most operators haven't seen much real-world impact from AI yet. Genuinely agentic systems — ones that can plan, act, and adapt rather than just crunch numbers — have only been production-ready since late 2025. Before that, the models were either too expensive (around £20 per task), too unreliable (going off-piste two times out of ten), or both.
That changed rapidly. Costs dropped from pounds to pennies. Reliability improved dramatically. If you're still working out what this means for your business, you're not behind. You're probably in the top half of the adoption curve. Your LinkedIn feed might suggest otherwise, but genuinely agentic AI is a very recent development.
Most AI tools in hospitality today are built on machine learning: forecasting demand, predicting revenue, pricing dynamically. These work well in stable environments. Hospitality rarely offers that.
Agentic AI works differently. Rather than pattern-matching against historical data, it creates a plan, executes it, learns from the results, and revises its approach, much closer to how a good operator actually thinks.
One of the most useful frameworks from the talk was a parallel between AI adoption and management development. Operators move through four stages:
Stage 1: Micromanagement. You give the AI a single, specific task, like finding the best person for a Thursday shift.
Stage 2: Task delegation with context. More background, more constraints, better output, but you're still directing every step.
Stage 3: Outcome delegation. You give the AI an objective like "build next week's roster balancing forecast, preferences, and labour cost" and a GM spends ten minutes reviewing what used to take four hours.
Stage 4: Proactive monitoring. The AI notices bookings are up 20% by Tuesday, cross-references a sunny weekend forecast, and recommends opening the terrace with a staffing plan attached.
Operators build dashboards to track what matters. But attention is finite, most people can actively monitor about five things at once. Dashboards multiply. Nobody checks them all. Problems resurface months later when focus shifts elsewhere.
An AI agent can monitor an entire estate, indefinitely, without losing focus. It can correlate food costs with staffing patterns, training completion, menu changes, and busyness levels, then flag a specific insight: "If these staff completed this training module, you could improve yields and reduce cost of goods sold." Then it scales it: "Want me to check whether this pattern exists across the rest of the estate?" The shift is from reactive problem-fixing to proactive improvement across the whole business.
Every operator knows what happens when a strong GM moves on. The small, instinctive calls about staffing, pacing, and priorities walk out the door with them.
When good decision-making is captured in systems rather than held in one person's head, new managers inherit it immediately. Operational knowledge compounds rather than resetting with every personnel change.
“That thing that we were planning to ship in 2027. We're shipping it now. That's not taking their job away. That's making them look like a superhero.”
The expected response to AI adoption is resistance. The reality has been different. AI handles the work that people know they should be doing but don't have time for. When it takes on those tasks, managers look better prepared in weekly reviews, have more time on the floor, and start pulling AI into other parts of their role.
"I don't think that many people got into hospitality saying, I can't wait to spend seven hours sitting in the back room writing a roster."
Everyone has a different capability profile. Some managers are natural coaches who struggle with spreadsheets. Others are analytically sharp but less confident on the floor. AI handles the areas where individuals are weaker, and lets them spend more time on the things they love and excel at.
Don't start by trying to buy an AI solution. Look at the parts of your business where more leverage would solve a meaningful problem. Where are your managers spending time on work that doesn't need to be done by a human? Where does knowledge disappear when people leave? Where are you monitoring things inconsistently?
Start there. The technology is ready. The entry point isn't technology, it's understanding where better decisions, made more consistently, would make the biggest difference to your operation.
AI is changing how the best hospitality operators run their businesses. But between the hype and the jargon, it's hard to know what's real, what's relevant, and where to start. This guide breaks it down. Written for hospitality leaders, not developers, it covers what AI actually is, what agentic AI means in practice, and how to apply it across scheduling, HR, performance, and multi-site operations.