Ask most people what an AI consultant does for a hospitality business and you will get one of two answers. A chatbot bolted onto the website, or a vague promise about the future of dining. Neither one pays your wage bill on a wet Tuesday in February.
The real job is quieter than the hype. It is also far more useful. It looks less like software than you would think. Rather more like standing in your kitchen at eleven on a Thursday morning, watching how the prep actually happens, asking why the order sheet gets written out twice.
Here is what it actually involves, in plain English, from a business that spent thirty five years in the trade before any of this.
It starts with the money leaks, not the software
A consultant worth hiring does not turn up with a product to sell. They walk your operation and find the one or two jobs quietly costing you most.
In a kitchen it is usually the same short list. Labour drifting over budget because the rota was built on a hunch. GP you cannot see until the month is closed, by which point it is too late to do anything about it. Stock you over-ordered because nobody had time to check what was already sitting in the walk-in. Paperwork eating the manager's evenings, which is how you lose good managers.
None of those are technology problems. They are time problems and visibility problems. AI is simply a tool that closes those gaps faster than a spreadsheet ever will.
The real skill is knowing which gap to close first, then being honest about the size of the prize. That order matters more than the tooling. Close the wrong gap first and you have spent money proving the technology works while the actual leak carries on quietly in the background.
A proper first pass is therefore closer to an audit than an installation. If you want to see what that looks like before you speak to anybody, we published the method as a free walkthrough in the bottleneck audit.
What the work actually looks like on the ground
Here is the part most articles skip. What does the consultant actually do all day.
The first visit is mostly watching and asking. Where does the day start. Who opens. What gets written down, on what, then typed up again later into something else entirely. Which supplier still phones instead of emailing. What happens when someone calls in sick at four o'clock on a Saturday. Which report does the accountant ask for every month that nobody enjoys producing.
The second pass is arithmetic. That is usually the part that changes the conversation. How many minutes does that job take. How many times a week. Who does it, at what hourly cost. Then multiply it out across a year.
Most owners are genuinely surprised by the total. Not because they are careless, but because the jobs are individually small. Four minutes does not feel like a problem. Four minutes, six times a day, across a full year, is a different conversation entirely.
Only after that does anything get built. What gets built is usually smaller and duller than people expect, which is generally a sign it is the right thing.
It builds around your operation, not a demo
Generic AI advice falls apart the moment it meets a real service.
The work that sticks is built on your covers, your suppliers, your menu, your team. In practice that might mean demand and labour forecasting tied to your own booking history, so you roster to the shift you will actually have rather than the one you feared. It might mean live menu costing that flags the dishes bleeding margin while supplier prices move week to week, instead of a costing exercise you did once in March then never revisited.
It might mean allergen and spec sheets generated in minutes rather than evenings. Replies to Google and TripAdvisor drafted in your voice, ready for you to read, edit, then approve, rather than sitting in a tab for a fortnight.
One or two jobs done properly beats a platform you will never fully use. Every kitchen has a piece of kit in the corner that seemed like a good idea at the time. Software is no different, except that software bills you every month whether you open it or not.
The detail on the specific tools, from labour forecasting through to menu costing, is laid out on our AI consultant for hospitality page.
It trains your team, then steps back
A tool nobody uses is worse than no tool at all, because you paid for it.
The point of good AI work is that your team owns it after the consultant has gone. That means hands-on training on the real tasks, in plain English, using your actual system rather than a slide deck, until it clicks. Chefs, front of house staff, managers, each learning the part that helps them specifically rather than sitting through a generic hour about artificial intelligence.
Then a follow-up a few weeks later, because the habit is what decides whether any of it survives past week one. Plenty of perfectly good systems have died quietly in the second month.
This is usually where the worry about jobs gets settled. It tends to get settled by the work rather than by anybody arguing about it. The tasks that go first are the ones nobody wanted in the first place. Retyping an order that already exists on an email. Chasing a spec sheet. Building next week's rota at the kitchen table on a Sunday night. Nobody went into hospitality to do those.
If you want that side handled properly for your team, our AI training work covers exactly that.
It tells you when AI is not the answer
This is the part the hype merchants skip entirely.
Sometimes the honest answer is that AI will not earn its keep on the problem in front of you. Your issue is a staffing one, or a menu one, or a supplier one, or simply that the site is too small to carry the overhead of another system. A consultant worth the fee will say so plainly.
You would far rather hear that in a free consultation than three invoices later.
There is a version of this business that involves selling every client the same platform regardless of fit. It is more profitable in the short run. It is also the reason a lot of owners have stopped answering the phone to consultants at all, which is a shame for everybody.
Why having worked the pass matters
TwoBridges.ai came out of thirty five years in commercial catering. Kitchens, front of house, wholesale food buying then selling, all of it long before any of the software.
That matters for a practical reason rather than a sentimental one. We speak GP, covers, section, wastage. You do not have to translate your business into technology language before the conversation can start. When somebody says the Tuesday night section is the problem, that lands without needing to be explained.
It also means a fairly low tolerance for nonsense. Anyone who has stood on a pass through a genuinely bad service develops a good instinct for the difference between something that works when it is quiet and something that works at eight o'clock on a Saturday. Most demos are the former. Most kitchens need the latter.
That is the whole point of the name. A two way bridge between your industry and the tools, so you get a straight answer rather than a sales pitch.
What it should cost you to find out
A fair question at this point is what any of this costs, because that is usually the thing standing between curiosity and a phone call.
The first conversation should cost you nothing but the hour. Any consultant who wants money before they have seen your operation is selling you something they decided on before they met you.
After that, the honest answer is that it depends on which job you are fixing, though the shape of it is predictable enough. A single well chosen job, costing, forecasting, the supplier paperwork, is a small piece of work with a return you can usually see inside a month or two. A full operation, several sites, integration with the systems you already run, is a bigger one. What should never happen is an open ended monthly fee attached to something nobody in your building understands well enough to question.
Ask any consultant three things before you commit. What exactly will exist at the end of this. Who owns it when you have gone. What happens if it does not work. The answers tell you almost everything. They also take about two minutes to get.
Where to start
If you are curious rather than convinced, that is the right place to be. It is a better starting point than enthusiasm, in our experience, because it asks better questions.
The first conversation is free. No pitch, no obligation. We will look at where the hours actually go in your operation, then tell you honestly where AI would pay back in your kitchen and where it would not. If it makes sense after that, we run a formal efficiency analysis and put real numbers against it.
If it does not make sense, you will hear that too. That costs you an hour and saves you rather more.