Something quietly broke in UK hospitality in the last two years. Not the food, not the service, not the desire people have to go out and eat somewhere wonderful. What broke was the economics.
From April 2026, the National Living Wage rose to £12.71 per hour. Employer National Insurance contributions increased from 13.8% to 15%, with the threshold dropped so that more of the wage bill is subject to contributions. Business rates relief — which had been sitting at 75% — has been reduced. Food costs are running 37% above 2020 levels. Energy costs remain around 70% higher than before 2022.
The numbers are stark. Official insolvency data shows that over 3,350 accommodation and food service businesses entered insolvency proceedings in the year to December 2025. That is not a correction. That is a structural crisis.
And yet consumer demand for hospitality remains. Premium venues maintain strong booking patterns. People still want to go out. The crisis is not on the demand side. It is entirely on the cost side — and specifically on the question of how those costs are structured.
That is exactly where this conversation needs to start. Not with technology for its own sake, but with a very specific question: which costs in your business could be handled by a system that never sleeps, never forgets, and scales without a salary?
There is a lot of noise around AI right now. Much of it is abstract. Agentic AI is a useful term, but only if we are clear about what it means in practice — particularly for a restaurant or bar running on tight margins with a team of six.
A chatbot responds to questions. Agentic AI acts on goals. The distinction matters enormously in a hospitality context.
A chatbot on your website can answer "what time do you open?" An agentic AI takes a booking request at 11pm on a Wednesday, checks real-time availability, applies your cancellation policy, notes the shellfish allergy the guest mentioned, assigns a table, sends a confirmation email, and adds the allergen to that guest's permanent profile — all without anyone on your team being involved, aware, or awake.
The agent does not just answer. It executes.
The shift that matters for hospitality operators: traditional software responds to instructions. Agentic AI works toward outcomes. The question is no longer "what can I automate?" It is "which outcomes in my business are currently achieved by paying someone to remember things, coordinate people, and prepare information — and how much of that could an agent handle?"
IDC, in its 2026 hospitality technology forecast, frames it this way: personalization in hospitality is shifting from "what offer should we send?" to "how should the experience adapt right now?" A restaurant that surfaces menu recommendations based on a guest's past visits and real-time kitchen inventory is not using technology as a novelty. It is using technology to deliver what good hospitality has always required — knowledge of the guest.
When hospitality operators talk about labour costs, they tend to focus on hourly rates — understandably, because those are the costs that appear on payroll. But the most expensive hours in a restaurant are rarely the ones spent in front of guests.
Consider what a manager in a 50-cover restaurant does in the hour before service:
That is approximately 60 to 90 minutes of management time per service. At a London management rate of £18.50 per hour including employer costs, that is £27 to £41 per service, or approximately £500 to £800 per month — just in pre-service preparation time.
That money is not going toward the food. It is not improving the guest experience. It is going toward the act of preparing information that already exists somewhere in a notebook, a spreadsheet, and three people's memories.
This is where agentic AI has its most immediate, measurable impact on a restaurant's cost structure. Not by reducing headcount. Not by replacing anyone. By doing the preparation work automatically, so that the hour before service is spent on hospitality — not on information retrieval.
There is a line of thinking in hospitality that technology threatens the guest relationship. That replacing a phone call with a chat booking, or a handwritten note with a digital profile, somehow erodes the warmth that makes a restaurant special.
This gets things exactly backwards.
The guest relationship is not threatened by technology that knows them. It is threatened by the current reality: that most restaurants cannot possibly know most of their guests, because the knowledge is fragmented, perishable, and exits the door every time a member of staff leaves.
Research from Bloom Intelligence, published in early 2026, puts the problem in stark terms: 70% of first-time restaurant guests leave and never return — silently, without complaint, simply because nothing in the experience gave them a reason to feel known. The same research shows that AI-driven engagement can recover 38% of at-risk guests automatically.
The paradox of the best meal you ever had: it probably felt personal. Not because of the food alone — because someone made you feel expected. The table was right. The server knew about the dietary requirement without making a fuss of it. Your occasion was acknowledged. That feeling was not magic. It was the result of someone having taken the time to know something about you before you arrived.
The problem is that restaurants cannot do this manually at scale. The team member who knew your preference left six months ago. The note that captured it is in a book no one reads. Agentic AI solves this — not by simulating warmth, but by making the information available to the people who can act on it.
As Chema Basterrechea, Chief Operations Officer at Radisson Hotel Group, put it: "Generative AI is shifting travel planning from traditional search into more conversational and agent-led environments. That makes building relationships with customers more strategic and important, not less."
The most valuable thing a restaurant can build right now is a guest intelligence asset that does not depend on any single employee's memory. Every allergen noted. Every preference captured. Every occasion remembered. Not in someone's head — in a system that is there at every service, for every team member, across every visit.
Independent UK restaurants are not short of technology options. The challenge in 2026 is not finding tools — it is deciding which tools actually improve the financial position of the business, and which simply add another subscription to the monthly cost base.
The framing that matters is not "what technology should I buy?" It is "what is currently costing me money that technology could handle better and for less?"
There are three categories of cost that agentic AI meaningfully addresses for an independent restaurant:
Most UK restaurants are paying separately for a booking system, a POS, and some form of guest management. These three subscriptions commonly total £400 to £700 per month — before a single staff hour is counted. A platform that replaces all three with a single, integrated subscription does not require a new budget line. It replaces existing ones.
Pre-service preparation, booking admin, follow-up communications, and manual reporting are all examples of management time spent on coordination rather than operation. At £18 to £20 per hour true cost, this is the most expensive category of work in the building. An agentic system that handles booking confirmations, generates pre-service briefings, and surfaces guest intelligence automatically returns real cash to the business.
No-shows cost the average 60-cover UK restaurant an estimated £5,000 to £10,000 per year in lost covers. An AI system that identifies high-risk bookings, automatically sends pre-confirmation requests, and manages waitlists in real time is not a marketing tool — it is a revenue recovery mechanism.
The technology is not an addition to costs. It is a restructuring of costs — shifting spend from coordination (which technology handles well) toward craft (which only your team can deliver).
Every serious conversation about AI in hospitality eventually reaches the same point: what happens to the people?
The answer, borne out by operators who have implemented AI systems, is consistent: AI does not replace the people who make hospitality what it is. It eliminates the administrative overhead that currently surrounds them.
QSR Magazine's 2026 operator survey captures the strategic direction clearly: "AI and automation should reduce cognitive load, not replace the human touch. When employees have fewer configuration tasks and fewer errors to troubleshoot, they can focus on hospitality — the very thing that still differentiates restaurants from delivery apps."
The head chef who is not spending Tuesday afternoon on a stock sheet is spending it on the menu. The floor manager who receives an AI-generated briefing rather than spending 90 minutes building one from scratch is using that time for the relationships that turn first-time guests into regulars. The owner who is not answering booking enquiries at 11pm is actually resting — and making better decisions the next morning.
The most labour-intensive parts of running a restaurant — the skill, the instinct, the human connection — are not what AI does. AI does the parts that should never have required human time in the first place.
Hanna AI has been running every service at Hotori restaurant in London since 2024. Not as a pilot. Not in parallel with another system. As the operating infrastructure for a real kitchen, a real team, and real guests.
In practice, agentic AI in a working restaurant means:
None of this replaces the food. None of it replaces the floor team or the kitchen. What it replaces is the scaffolding of coordination that currently surrounds them — and which, in a cost environment as severe as the one UK hospitality faces in 2026, can no longer be sustained at its current price.
The hospitality businesses that will survive the current cost environment are not the ones that cut the most. They are the ones that redirect the most — moving budget from what machines can do to what only people can do.
Food costs rising 37% above pre-pandemic levels are not something technology fixes. Energy costs 70% above 2022 levels are not something technology fixes. But the 30 to 40% of labour budget currently spent on coordination, administration, and preparation — that is exactly where agentic AI changes the economics.
The question is not whether AI belongs in hospitality. The evidence from 2026 is unambiguous: it does, it works, and the operators who move first capture the margin advantage while others are still debating the philosophy.
The question is which technology actually delivers the return — and whether it is structured as a tool that adds to your cost base, or a system that restructures it.
Hanna AI is live at Hotori restaurant in London. AI booking, full POS, pre-service intelligence and guest memory — one flat monthly fee, zero per-cover commission. The founding rate — £300/month Core or £225/month Intelligence — is available to the first 20 venues. Book a demo to see the system running in a real restaurant.
Book a demo →