AI in Hospitality: Where It Helps Restaurants and Hotels Today

Hand-drawn sketch: a wall splits the back office from front of house. In the back office the AI works through a checked rota and sorted paperwork. Out front, staff serve guests at a table

AI in hospitality helps in two places today. On the guest side it prices rooms, answers booking questions and personalises offers. On the people side it answers staff questions, turns policies into training, drafts team updates, checks rotas and shows which sites are slipping. What works right now is AI that answers, drafts, checks and finds, with a manager making the call. What is still mostly hype is AI that runs a site on its own.

This guide is for people running hospitality groups with five or more sites: Heads of People, Ops Directors, Area Managers and founders. It covers where AI for hotels and restaurants is real today, what is still a pitch deck, the risks (data protection, accuracy and trust), the EU AI Act basics, and a plan to start. We build AI for hospitality teams, so we see a lot of it. We have tried to keep the sales pitch to one section near the end.

AI in hospitality is not one thing. Pick the job first, the tool second. Want to see AI do the people admin across your sites? Book a demo.

Where AI in Hospitality Stands Today

Hospitality is a late adopter, and the numbers say so.

  • Three in ten UK firms use AI. 30% of UK businesses used at least one AI technology in September 2026, rising to 38% of firms with 10 or more employees and 51% of firms with 250 or more (ONS, Business insights and impact on the UK economy, 8 October 2026).
  • Hospitality sits near the bottom. In June 2026, about 21% of accommodation and food service businesses used AI, 10th of 13 industries. Information and communication was at 58% (ONS, Artificial intelligence in UK businesses: 2023 to 2026).
  • Most of it is free tools. In the same ONS article, 54% of hospitality businesses’ AI use came through free-to-use software, against 13% through software they bought.

That last point matters more than the headline number. In most groups, AI already arrived through the back door: a GM writing the rota email in a free chatbot, a head chef pasting a supplier spec into it, an HR manager asking it to rewrite a policy. The question is not whether your teams use AI. It is whether they use it on your terms.

AI on the Guest Side: Revenue, Chatbots and Personalisation

The guest side got AI first because revenue is easy to count. Our CEO wrote about why in hospitality is pointing AI at its guests. Here is what that looks like in practice.

Revenue management and demand forecasting

Hotels have used pricing software that reacts to demand for years. Newer systems use machine learning to forecast demand from bookings, events and seasonality, and suggest a rate. Restaurants use the same idea to forecast covers, which feeds purchasing, prep and how many people to put on a shift. Real today, with one caveat: a forecast is only as good as the history behind it, so a new site or a new concept starts blind.

Chatbots and guest messaging

Booking questions, opening hours, allergens, late check-out, parking. A chatbot on the website or in a messaging app handles the repetitive ones around the clock and hands the rest to a person. Real today, as long as it answers from your own approved information. A chatbot that improvises on refunds or allergens is a liability, not a feature (more on that under risks).

Personalisation and marketing

Offers based on past visits, emails written for a segment, menu recommendations, upsells at booking. Real, but uneven. It works where you have clean guest data and a loyalty or booking system that holds it. For many restaurant groups that data is thin, and the personalisation is mostly better-written marketing.

Reviews and guest feedback

Summarising hundreds of reviews into themes per site, drafting replies for a manager to send. Real today and an easy win, because a human still reads before anything goes public.

GUEST SIDEPEOPLE SIDEPRICING, BOOKING CHATPERSONAL OFFERSONBOARDING, TRAININGCOMMS, ROTAS, QUESTIONSAI

AI on the People Side: Where Multi-Site Groups Get Hours Back

The people side is where multi-site groups lose the most hours. Every site repeats the same admin: set up the new starter, chase the training, answer the same question about holiday, copy the update into 20 group chats, type up the rota. It works at three sites. At five it breaks, and the Head of People becomes the bottleneck.

This is where AI for hotels and restaurants is quietly most useful, because the work is repetitive, text-heavy and rule-based (we talk it through in the podcast episode AI in hospitality: invest in people). Six jobs stand out.

Onboarding new starters

A new starter needs an account, the right groups, the right courses for their role and site, and their first tasks before day one. AI can prepare all of that from one sentence (“8 new starters at Soho, 5 FOH and 3 kitchen”) and a manager checks it before it goes. See AI onboarding.

Training and compliance

Turning a policy nobody reads into short lessons and a quiz used to take an L&D manager days. AI drafts it in minutes from your own document. The real value comes after: asking “which sites are behind on allergen training?” and getting names, not a spreadsheet hunt. See AI training and compliance.

Internal comms

A menu change, a new process, a reminder about the bank holiday rota. AI drafts the update in your voice and picks the right roles and sites, so the Brighton kitchen does not get the Manchester bar’s news. See AI internal comms.

Staff questions and the HR helpdesk

“How do I call in sick?” “How much holiday do I have left?” “What is the allergen process for the new dish?” These questions land on managers, often on their day off. An AI helpdesk answers from your own handbooks, links the page it used and stays quiet when the handbook does not cover it. See AI HR helpdesk.

Scheduling and rotas

This is where the hype is loudest, so be precise. What works today: turning a photo of a paper rota into a draft, filling open shifts you ask it to fill, and checking a rota against your rules before it is published. What it should not do is plan your rota on its own and publish it. A rota is a set of judgement calls about people, and a manager should make them. See AI rota checks.

People insights

Which site is slipping on training? Where are new starters not finishing onboarding? Which team has gone quiet? AI can answer those questions in plain words from your own data. That matters, because engagement predicts who stays. In our Hospitality Retention Benchmark 2026 of 28,002 employees, the most engaged third left 69% less often than the least engaged third. See AI people insights.

AI for Hotels and Restaurants: What’s Real and What’s Hype

A quick sense check for the next vendor pitch.

ClaimReal today?What to look for
AI suggests room rates from demandYesYour history, your override
AI forecasts covers for a siteYes, with historyAccuracy per site, not group average
A chatbot answers guest questionsYesAnswers only from your approved information
AI answers staff questions from the handbookYesA link to the source on every answer
AI turns a policy into a course and quizYesYou edit and approve before it is assigned
AI drafts and targets team updatesYesNothing posts without a manager’s OK
AI checks a rota against your rulesYesIt flags, you decide
AI plans and publishes your rota on its ownNoTreat “fully automatic” with suspicion
AI agents run a site without supervisionNoAsk what it did last week without a human
AI decides who gets hired or let goShould notHigh-risk under the EU AI Act, and a trust killer

The pattern is simple. Where AI drafts, checks or finds and a person decides, it is real and useful. Where it decides on its own about people, it is either not ready or not wise.

The Risks: Data Protection, Accuracy and Trust

AI on the people side touches personal data and people’s livelihoods. Three risks deserve a plan.

Data protection

Staff data is personal data, so UK GDPR (or EU GDPR) applies to any AI that uses it. The rule that matters most for HR: people have rights around significant decisions made solely by automated means. In the EU that is Article 22 of the GDPR. In the UK, the Data (Use and Access) Act 2025 replaced it with Articles 22A to 22D, in force since February 2026: a decision counts as solely automated if there is no meaningful human involvement, and people get safeguards such as information about the decision, the right to make representations, human intervention and the right to contest it (Article 22C).

In practice: keep a real human in every decision about hiring, pay, shifts and discipline. Ask any vendor where your data is stored, whether it trains their models, and how roles limit what the AI can see. The ICO’s guidance on AI and data protection is the UK starting point. Free consumer chatbots are where most of the risk sits today, because staff paste in things they should not. A short GDPR training course for staff helps with exactly that.

The EU AI Act basics

If you run sites in the EU, the EU AI Act (Regulation (EU) 2024/1689) applies there. UK-only groups are outside it, but UK GDPR still applies. The basics, from official EU sources:

  • In force since 1 August 2024, with obligations phased in (European Commission, AI Act).
  • Since 2 February 2025: banned practices, including AI that infers the emotions of people at work from biometric data such as faces or voices (Article 5, Article 3). Also a duty to take measures to support AI literacy among staff who use AI (Article 4).
  • Transparency: AI that talks to people must make clear it is AI, unless that is obvious (Article 50).
  • High-risk HR uses: AI for recruitment and selection, for decisions on terms of work, promotion or termination, for allocating tasks based on individual behaviour or personal traits, and for monitoring and evaluating performance and behaviour is classed as high-risk (Annex III, point 4). After the 2026 amendment (Regulation (EU) 2026/1744), those obligations apply from 2 December 2027.

This is a summary, not legal advice. The takeaway for a hospitality group: answering questions, building training and drafting comms are low-risk jobs. Scoring, ranking or deciding about people is where the law gets strict.

Accuracy

AI can sound sure and still be wrong. The case people cite is Moffatt v. Air Canada: an airline’s website chatbot gave a customer wrong information about a fare refund, and a Canadian tribunal held the airline responsible, noting it is responsible for all the information on its website (Civil Resolution Tribunal, 2024 BCCRT 149). Swap refunds for allergens or sick pay and the stakes go up.

The fix is boring and it works: let AI answer only from your own approved sources, show the source with every answer, keep a human approval on anything that goes out to teams, and keep the sources current.

People’s trust

Staff will ask two questions about any AI at work: is it watching me, and will it replace me? Answer both before you launch. Say what it does and what it does not do. Keep anything that touches mood or sentiment anonymised and at team level, never per person. In the EU, check local rules on consulting employee representatives. Trust is the asset: a team that trusts the tool asks it the awkward questions they would never ask their manager.

AI DRAFTSYOU DECIDETHEN IT GOES OUT

How to Start With AI in a Multi-Site Group

Start with one job, not a strategy deck. Here is a plan that fits a quarter.

  1. Pick one job with volume. Staff questions, compliance training or onboarding are good first picks: high repetition, clear right answers, low risk. Avoid anything that decides about people.
  2. Fix the source first. AI answers from what you give it. If your handbook is a 50-page PDF from 2021, update the five policies people ask about most before you switch anything on.
  3. Set the guardrails. A manager approves anything that goes out. Roles limit what the AI can see. Write down what it is for, what it is not for, and who owns it.
  4. Tell the team. A short post: what it is, what it is not, where the data goes, who to ask. Make it the opposite of a surprise.
  5. Pilot at two sites for six weeks. One strong site and one that struggles. Measure questions answered, hours saved for managers, training completion and onboarding done before day one.
  6. Roll out site by site. Keep what moved the numbers, drop what did not, then add the next job.

Questions worth asking any vendor:

  • Does it answer only from our own content, and show the source?
  • Does a manager approve before anything is posted, assigned or changed?
  • Can it see only what each person’s role allows?
  • Where is our data stored, and is it used to train models?
  • Which jobs is it live for today, and which are on a roadmap?

How All Gravy’s AI Assistant Fits

All Gravy is the AI employee app for hospitality, used by 350+ brands across 4,000+ locations. The AI Assistant is the AI inside it, built for the people side. It works on the handbooks, courses, sites, roles and people already in the app, and nothing is posted, assigned or changed until a manager approves.

It does six jobs today:

  • Onboarding: new starters set up with the right tasks, courses and groups before their first shift.
  • Training and compliance: your policy becomes a course and quiz, and you see who is behind.
  • Internal comms: the update drafted in your voice and sent to the right roles and sites.
  • HR helpdesk: staff questions answered from your own handbooks, with the page it used.
  • Rota checks: a paper rota becomes a rules-checked draft in Scheduling, and you make the call.
  • People insights: which sites and people are behind, in plain words.

Want to see it on your own handbooks and sites? Book a demo.

Frequently Asked Questions

What is AI in hospitality?

AI in hospitality is software that predicts, writes, answers or checks things for restaurants, hotels, pubs and bars. On the guest side it prices rooms, answers booking questions and personalises offers. On the people side it answers staff questions, builds training, drafts team updates, checks rotas and shows which sites need attention.

How is AI used in restaurants?

Restaurants use AI to forecast covers, answer booking and menu questions, write marketing, summarise reviews and, on the people side, to answer staff questions from the handbook, turn policies into short courses, draft updates for the right sites and check rotas against the rules before they go out.

How is AI used in hotels?

Hotels use AI most for room pricing and demand forecasting, guest messaging before and during the stay, and personalised offers. Behind the scenes the same people-side jobs apply as in restaurants: onboarding seasonal staff, training, internal comms and staff questions across departments.

Will AI replace hospitality staff?

Not the people who serve guests. What AI takes over is admin: the same answer for the tenth time, copying a policy into a course, typing up a paper rota. That gives managers time back for their teams. Hospitality is still a people business. Our 5 myths about AI at work goes deeper.

Yes, with care. UK GDPR and EU GDPR give people rights around significant decisions made solely by automated means, so keep a human in every decision about hiring, pay, shifts and discipline. In the EU, the AI Act treats AI used for recruitment, promotion, termination, task allocation based on personal traits and performance monitoring as high-risk, with those obligations applying from 2 December 2027.

How should a multi-site hospitality group start with AI?

Pick one high-volume people job, such as staff questions or compliance training. Get the source material right first, set the guardrails (a manager approves, roles limit what AI sees), pilot at two sites for six weeks, measure hours saved and completion rates, then roll out site by site.

Where is this heading? Our CEO on AI employees that own a whole job, and what HR does with the time it gets back.

Take your people and culture to the next level

  • 350+ hospitality brands
  • 4,000+ locations
  • 4.7 on G2
The All Gravy app on a phone: feed with the channels For you, General, Customer love and Office