Psst… I’ve launched Adapt and Create, it explores thinking clearly, learning deeply and adapting to change. Edition two arrives Saturday.
Hi Adopter,
Before you raise a menu price, check when someone last updated the costs behind it. A spreadsheet can calculate every row correctly while using an old supplier price or the wrong delivery fee.
That’s a useful job to test with AI: compare the records, show what disagrees, and give your manager something they can check.
This week’s case starts with an owner running three restaurants. Then we’ll use a fictional menu item to see what that comparison should catch. You can try it with the supplied records before bringing in your own.
What Taylor checked first
Ted and Tami Taylor run three co-branded Schlotzsky’s and Cinnabon restaurants in New Mexico. In OpenAI’s October account, Ted reports cutting annual pricing analysis from five days to one or two hours with ChatGPT.
He first compared its pricing work with his own manual analysis. Later, when he tried scheduling at one location, the manager caught missing rules about adding staff during busy periods. They revised the instructions. The manager still reviews and enters the schedule.
Those details are more useful than the headline time saving. Taylor supplied business knowledge and checked the work before using it. OpenAI published the account, and the reported timings aren’t independently audited. Setup costs and error rates aren’t disclosed.
An earlier profile adds something worth considering: a detailed competitor-price comparison would otherwise have been skipped. The benefit can be getting a useful analysis done at all.
For your own restaurant, choose one item you sell in-store and through delivery. Bring its current supplier invoice, recipe, and delivery-fee agreement. Add the menu prices and the costing sheet you already use.
Ask the assistant to compare those records before calculating anything. Which prices are current? Does the recipe use pounds or ounces? Does the fee in the sheet match the agreement?
The member exercise includes the fictional records, two planted mistakes, and a checked answer. You’ll see what the assistant should flag, what your team must verify, and when the spreadsheet you already have is enough.
