The restaurant

Ace Pizza is Rachel Jones's restaurant, open just over a year in London after running as a pop-up and a residency before that. Tom Harrison came to it from fintech and runs operations.

That background shows in how they work. Recipe costings, gross profit calculators, wage percentage databases, all built in-house rather than handed to an accountant at month end.

We didn't want to be reliant on accountants. We wanted to use the data we had and do something with it from the off. Hospitality is difficult. You've got to be looking at everything daily and weekly, not monthly.

What was not working

The ambition was daily. The stocktake was anything but.

Drinks ran to about two hours. Food, at roughly three quarters of their stock, took four to six, because so much of it has to be weighed. A manual count on a printed spreadsheet, then typed into a CSV, then matched against cost holdings.

And the count itself was only half of it.

We might do a two-hour manual stocktake, but the review, the data entry, and pulling those numbers together for accounts is considerably more time on top.

How it runs now

Tom ran a bar count himself recently, handing over to a new management team. Filming, processing and a simple review came to 45 minutes end to end. Against two hours of counting plus everything that followed it, that is at least half the time on the drinks side, and the data entry is gone rather than reduced.

The review keeps getting shorter as the system learns how they film and how they talk.

The AI has got more used to our verbal instructions and our videos. I've noticed month on month how much that review reduces. I'm trusting more what the system's finding and what it's correcting naturally.

The kitchen is the harder problem, and Ace is a good test of it. Batch preparations, toppings that arrive in one pack and end up in a four-litre container, tens of anchovies, and a great deal that has to go on the scales in grams. It is a two-person job.

What changed

Ace moved through the system deliberately. Inventory first, then procurement and receiving stock, then the menu integration and Lightspeed. That last step is where the numbers start answering questions rather than just recording them: variance, and whether a pizza is being over-portioned.

The fact that it links to Lightspeed is a real positive for us. It takes that connection away. We don't have to apply that to something separate.

The costing side turned out to matter to people outside the management team.

Our bar supervisor is working on a new cocktail. To open the app, dial in those preparations and ingredients, and get an idea of whether his spec is going to make us profit, is really cool.

What they would tell another operator

In hospitality people are often scared to change, because we're so busy. It's hard to make change in something that's constantly active and reactive. I know I would put up walls to new software because I haven't got the time to implement it.

The onboarding has always been really supportive. Working with a startup, knowing you can have input on feature requests, and that if they're worthwhile they'll get implemented, has been a real positive for us.

Dip your toe into the inventory element. Over time, even if it's just on a stock level, you will save time. And the more you use the system, the more you'll save money as well.

Photography: Nic Crilly-Hargrave