A table orders the special. They wait. Fifteen minutes later the server comes back: sorry, we actually 86'd that an hour ago. Now they're annoyed, they're rushed into a second choice they didn't want, and the whole visit took a step backward — over a dish you already knew you were out of. It happens with the keg that blew mid-shift, the well tequila you ran dry, the fish that sold out at eight. The information existed the moment it ran out; it just never made it to the person ordering. An AI ordering assistant closes that gap — it reflects what's actually available right now, so the order nobody can fill never gets taken in the first place.
A printed menu doesn't know what's out — and neither does most ordering tech
The menu on the table, the QR-code menu, the static order screen — they all show the same thing at 9pm that they showed at open, whether or not you've got it. So the customer orders something that's gone, and the bad news gets delivered late, by a human, after they've already gotten their hopes up. The problem was never that you ran out — running out happens every single night. The problem is the lag between you knowing and the customer knowing, and that lag is where the soured order lives.
Mark it 86'd once — and it's gone everywhere instantly
When you run out, you flag it — the keg, the special, the well gin — and the assistant stops offering it, immediately, to every seat. No one can order what you can't make, because it's simply not on the table as an option anymore. When it comes back, you flip it back on. It reflects the availability you set — it's not magically tracking your inventory or counting bottles for you; you tell it what's out, and it makes sure that knowledge reaches every customer the instant you do, instead of one apologetic table at a time. That's the honest version: not an inventory system, a real-time menu that actually tells the truth about what's available.
The dead-end order becomes a save
Here's the part that turns a defensive feature into a revenue one. When someone reaches for something that's out, the assistant doesn't just say no — it offers the closest thing you'd want them to have instead. Out of the special? Here's the dish the kitchen would point you to. Blew the keg of the local IPA? Here's the other one on tap people love. A flat “we're out” loses the order. A “we're out of that, but this is the one regulars get when it's gone” keeps it — and often trades the customer up to something they end up liking more. The substitution suggestion is yours to define; the assistant just delivers it at the exact moment it matters, when they're deciding.
The order lands accurate, and your floor stops running interference
Two things stop happening once availability is built in. Servers stop being the bearers of bad news — no more walking back to a table to undo an order, no more renegotiating a second choice under time pressure. And the order that reaches your staff is one you can actually fill, every time, so the accuracy the whole ordering system depends on holds up instead of getting undercut by a menu that's out of date. The bottleneck you remove isn't just taking the order — it's all the cleanup when the order was for something that wasn't there.
What you need to stand one up
The inputs are your menu and a simple way for your staff to flag what's 86'd in the moment — plus, if you want the save, the substitution you'd suggest for the items that go out most. Hand that over and the availability controls come built in. Done for you, not do-it-yourself.
Running out is going to happen every night you're busy — that's not the problem worth solving. The problem is the customer finding out the hard way, fifteen minutes in. This makes what's out of stock known the instant it's out, turns the dead-ends into saves, and keeps every order one you can actually make.