Three days ago, on 1 July, VAT on food and catering services in Ireland dropped from 13.5% back to 9%. It was announced in Budget 2026 last October, the sector fought years for it, and there's no announced expiry this time.
So why does this feel like the right week to talk about fixing leaks rather than celebrating?
Because the Restaurants Association of Ireland, the body that campaigned hardest for the cut, welcomed it with a warning attached: the labour cost increases that arrived in January 2026 would eat much of the benefit. The people who wanted this most are telling you, on the record, not to spend it twice.
We run Resk (formerly Elyx), an AI operator platform for restaurants, so you know where we stand. But the argument here doesn't need you to buy anything to be true.
A reprieve, not a rescue
Think about what the VAT cut actually is. It's margin handed back by policy. You didn't earn it through better buying or tighter service, and you can't defend it, because the same policy machine can take it away in a future budget. It went up before. Owners who traded through 2023 don't need reminding.
Meanwhile the January labour increases are already in your payroll, compounding every fortnight. Rent didn't move. Energy didn't move. So the honest frame for this week is: you've been given a window, not a floor. Some percentage of the pressure just lifted, temporarily, by someone else's decision. The question is what you do inside that window.
There are two options. Let the extra margin quietly get absorbed into the same cost stack, which is what happens by default. Or use the breathing room to fix the leaks that are yours to fix, so that when the next cost rise lands, and the RAI has told you the schedule, you're running a tighter operation than you were in June.
Kill the 90% myth first
Whenever restaurant survival comes up, someone repeats the line that 90% of restaurants fail in their first year. It's folklore, and it's wrong.
Ohio State University research went looking for the real number, and the study's actual headline was that failure is far lower than commonly assumed. The real figure: roughly 59% of restaurants fail within their first three years, around 61% for independents.
Read that honestly and two things are true at once. The doom number people quote at you is a myth, and you should say so when you hear it. And the corrected number is still brutal. Six in ten gone inside three years is not an industry where you can afford to leak.
Here's the useful part. The research-backed drivers of closure are a mixed list: food costs, energy, wages, VAT, interest rates, labour shortages, delivery commissions, poor cashflow visibility, owner burnout. Look at that list again and sort it into two piles. The first pile, the macro pile, you cannot control. The second pile, the operational pile, you can. Missed demand, weak cashflow visibility, the burnout that comes from being the only person watching everything. Those are the ones that decide which side of the 59% you land on, because they're the only ones you get a vote on.
What an AI operator actually is
The phrase sounds abstract, so let's drop it into three moments from a real service week instead.
It's 7:40pm on a Friday. The room is full, the phone rings, and nobody can take it. Before this year, that call rang out, and the guest booked somewhere else within a minute. With an AI operator, the call is answered, the intent is captured, the diary is checked against your rules, and the table is held. Not a voicemail. A booking. The owner sees it in the morning brief with the guest's name attached.
It's Thursday afternoon and money is waiting in payment links. Two deposits for Saturday's large parties were never paid, and a payment link from a private hire failed silently on Tuesday. Nobody noticed because noticing was somebody's twelfth job. The operator notices, sends the chase, and hands the manager a clean list before service. If Saturday's eight-top no-shows, the deposit means it didn't cost you the full table.
It's Monday and a supplier price change is eating margin. The invoice came in higher on a line that touches four dishes. In most restaurants, that's discovered at the end of the quarter, if ever. The operator links the invoice change to the menu items it affects and drafts the decision for the owner: reprice, renegotiate, or re-spec the dish. The owner decides. The point is that the decision arrives in July, not October.
That's the whole idea. Systems you already have record these things. An operator acts on them, then compresses the week into one daily brief so the owner can run the restaurant instead of chasing it.
What it refuses to pretend to fix
We hold a hard line here because our industry doesn't always. An AI operator does not fix your rent. It does not fix energy prices, the VAT rate or the wage floor. Those are set in budgets and negotiations no software touches, and any vendor implying otherwise is lying to you at a vulnerable moment.
What it fixes is the controllable list: the missed calls, the unpaid deposits, the failed payment links, the no-shows, the margin drift after supplier changes, the roster that doesn't match the bookings, and the cashflow blindness that turns every month-end into a surprise. On a bad year, closing those leaks is the difference between a hard year and a final one.
Why now, specifically
Because the window is open and it has a shape. The VAT cut took effect three days ago. The labour increases that will absorb much of it are already running. Nobody knows how long the 9% rate survives future budgets, and the last cycle says don't bet the lease on it.
The owners who come out of this window stronger will be the ones who used the reprieve to fix what was theirs to fix. That work takes weeks, not quarters, but it doesn't start by itself.
Start with the number. The free leak audit at resk.ai/leak-audit takes five quick answers and gives you one clear number: what the leaks you control are costing you. If the number is small, you've lost three minutes. If it isn't, book a demo and start a 30-day trial while the window is still open.