Food cost this week, explained by your AI

"What was food cost last week?" sounds like a one-number question. It isn't. A food cost percentage is only as good as what's behind it: which items have recipes, what the ingredients cost, what got thrown out. The report gives you the number. Usually nobody explains what's behind it.

When Mays is connected to your AI app, you can ask for the number and for what's behind it, in the same message.

Part of our series Run your restaurant with your AI: one real situation per post, with the words you type and what happens in Mays. The prompts and results are illustrative examples.

What the owner types

The owner of Harbor Lane Café asks on Monday morning:

"What was my food cost % last week at Midtown and Riverside? Explain anything I should be careful about."

What the AI does in Mays

It asks Mays for food cost and prime cost for last week, per location and in total. That's the same figure as Insights → Prime Cost in Mays, with more around it. Each location comes back with:

• Net sales from the POS orders Mays receives (Square or Union).
• Theoretical food cost in dollars and as a % of sales: what the items you sold should have used, from their recipes and your ingredient prices.
• Recipe coverage: the share of item sales that's mapped to a recipe at that location, plus the best-selling items that still have no recipe.
• Purchases: what you bought on posted invoices in the same days. That's a measured number, but it's what you bought, not what you used.
• Food cost by cost class (food, beverage, supply) and the ingredient categories that cost the most.
• The Waste Log: what was thrown out, valued at the item's cost when it was logged, by reason and by item.
• Labor cost, labor % and prime cost (labor plus food cost), but only if you can open Insights, can see pay in Mays, and chose to include pay when you connected.

Each figure comes with a note on how it's measured, and the AI is told never to fill in a number that's hidden.

What a good answer sounds like

Here's the kind of explanation the AI can give from that data (illustrative numbers):

"Midtown: food cost 27.9% of $18,420 net sales. Only 64% of item sales are mapped to recipes, so the real figure is probably higher. Your top unmapped items are the breakfast burrito and the cold brew. Riverside: 31.2%, with 92% mapped. Waste at Riverside was $310, mostly 'expired', led by croissants."

Two things in that answer you wouldn't get from the bare percentage:

1. Midtown's number looks better than it is. Items without a recipe add sales but no cost, so low coverage pulls the percentage down. Mays flags any location under 80% coverage.
2. Where to act first. Mapping two best sellers to recipes fixes more of the number than any other change, and the waste tells you which bake to cut.

What "theoretical" means, in one paragraph

Mays doesn't weigh your walk-in. Theoretical food cost multiplies what the POS says you sold by what each recipe says it uses, at what your ingredients cost in Mays (the vendor price, or an imported cost). If the recipe is right and nothing is wasted, stolen or over-portioned, that's what you spent. The gap between that and reality is what waste logs, counts and portion checks are for. Ingredient prices come from vendor prices in Mays, so after a supplier raises a price, food cost catches up when the new price is recorded, for example when you post the invoice.

Ask the follow-up

The useful part is the second question:

• "Show it week by week for the last 6 weeks." Mays adds weekly rows (up to 92 days at a time).
• "Which categories drive the cost at Riverside?"
• "How much did we waste this month, and why?"
• "Compare last week with the same week last year." (That's a comparison; Mays answers it too.)

What it replaces

Who can ask

Anyone who can open Insights, or Inventory reports when the Inventory module is on. Waste needs access to the Inventory Waste Log. Labor and prime cost need Insights, pay access in Mays and a connection that includes pay. This is a read-only question: no special setting is needed beyond connecting.

What Mays doesn't do here

• No "actual" food cost from inventory counts. Opening inventory plus purchases minus closing inventory needs counts with dollar values, and Mays counts store quantities only. The answer says so instead of guessing.
• It doesn't cost items you haven't mapped. Map POS items to recipes in Inventory → POS mapping to raise coverage.
• It doesn't invent labor numbers. Without pay access, labor and prime cost are left out, not shown as zero.

Try it this Monday

1. Connect Mays to ChatGPT, Claude or Grok from mays.app/ai (ChatGPT, Claude, Grok). See Connect Mays to ChatGPT, Claude or Grok for what you approve, including pay.
2. Ask for last week's food cost, and ask the AI to explain the caveats.
3. Map the unmapped best sellers first. Recipes and plate costing and Prime cost show how the numbers are built.

Try Mays free for 30 days, with no credit card, and connect it to the AI you already use.