Use cases

Set up a new café in Mays from one message in ChatGPT or Claude

Mays TeamOctober 12, 2026
Mays cover card: Set up a new café in Mays from one message in ChatGPT or Claude — Staff, hours and positions in, a working store out

New software usually starts with an afternoon of forms. Add the location, add each position, add each person one at a time, build the first week, set up the opening checklist, set the clock-in radius, and then remember what you forgot. Most of that information already exists somewhere: a staff spreadsheet, a note with your hours, a text thread with your opening crew.

With Mays connected to ChatGPT or Claude, you hand all of it over in one message, and Mays sets up the store from it, after showing you exactly what it will do.

This post is part of our series Run your restaurant with your AI.

The situation

You've just signed up for Mays for Harbor Lane Café. You have one location, Midtown, eight people, three positions and opening hours you know by heart.

Before: you copy each name and phone number from your spreadsheet into Mays, one form at a time, then build the week shift by shift, then type the checklist items in from the paper version taped to the wall.

Now: you connect Mays as the admin and paste everything into the chat:

Here's my café, Harbor Lane Café, Midtown location: 410 Harbor Lane, on Eastern time. Positions: barista, line cook, cashier. Here's my staff list with phone numbers and positions [pasted list]. Maya is the manager; her email is in the list. Weekdays we open 6:30 to 2:30 with two baristas and a cook; weekends Jordan does the line 7 to 3. Add an opening checklist: espresso machine on, pastry case stocked, register counted, floors checked. Clock-in radius 200 meters. Set everything up in Mays so my team can start.

Step 1: Mays checks the plan, and changes nothing

The AI turns what you sent into one plan and asks Mays to check it (plan_store_setup). Nothing is created yet. Mays answers with one preview:

  • how many locations, positions, people, shifts a week, checklists and settings it would add, with a table for each;
  • what's already in Mays and will be kept as it is (your first location, for example, created when you signed up);
  • the invites that will go out: a text for each employee with a phone number, an email for each manager;
  • every problem to fix first: a location without a time zone, two people with the same phone, a position mentioned for someone but never described.

The AI asks you about the problems ("Two people in the list share a phone number. Which one is Jordan's?") and checks again. Nothing can be applied until the plan is clean. Mays doesn't guess a time zone from an address; if it isn't clear, the AI asks you.

Step 2: one yes sets it all up

When the preview is clean, the AI shows it to you and waits. Your yes runs the setup (apply_store_setup) with exactly the plan you checked. Mays creates everything in order: locations, departments, positions, people (each gets the usual Mays invite), the weekly pattern, checklists, vendors and inventory items if you gave any, and settings such as the clock-in radius.

If something fails halfway, Mays reports exactly what it created and what didn't work. Fix the cause and say "Try the setup again." What already exists is skipped, so only the missing pieces are added and nobody is invited twice.

Step 3: the first week, and the team's welcome

The weekly pattern you described is saved, not scheduled. Three short follow-ups finish the job:

Put the weekly pattern on next week and publish it.

The AI creates next week's shifts from the pattern as drafts (apply_weekly_pattern), then publishes the week (publish_schedule_week). Each step has its own preview: the shifts it adds, then who will be notified.

Send the team their welcome.

Everyone who was invited but hasn't signed in yet gets the Mays welcome again: a text with the app download link for employees, the invite email for managers (send_team_welcome). The preview shows how many people, the exact message and who's skipped (already signed in, invited in the last 15 minutes, no phone or email on file, or replied STOP).

Is everything set up?

You get a checklist (get_onboarding_progress): locations and positions, who's invited, signed in and has the app installed, whether this week and next are published, checklists, time clock, pay period and time-off settings. Counts only, no names.

What you'll see along the way

Everything happens in the chat, as words and, in ChatGPT and Claude, Mays cards. People appear as first name and last initial. Phone numbers you pasted go to Mays so it can send invites, but the AI's answers never show them back; the preview lists invites masked.

What Mays doesn't do here

  • It doesn't invent your setup. It only builds what you told it. A missing piece is a problem in the preview, not a guess.
  • It doesn't publish anything you didn't ask for. The pattern is saved; the schedule is published only when you say so.
  • It doesn't replace your judgment on pay. Wages go in only if your connection includes pay. Without it, people are added without a wage, and you can set pay in Mays later.

Before you start

  • Connect as the admin. mays.app/ai has the setup for ChatGPT and Claude. New to Mays? You can create your account while you connect. See Connecting Mays to your AI app.
  • Allow changes when you connect. Right after sign-up, the Mays approval page offers you, as the admin, a box that turns on Allow AI apps to take actions for your organization and lets this AI app make changes. Without it, the connection can only read. Every change still shows you a preview and waits for your yes.

Already have a spreadsheet of your team? Import it with your AI. Opening a second café later? See open your second location by describing the first.

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