How a Columbus restaurant owner can use AI for schedules, prep lists and supplier emails

It is Sunday night. The kitchen is clean, the register is counted, and you are at the corner table with a cold coffee rebuilding next week's shift schedule from scratch. There is a prep list half scribbled on a napkin, two suppliers have not been answered since Thursday, and opening is at ten. A general AI assistant such as ChatGPT or Claude can draft all three of those things in minutes, from a few clear prompts, with no restaurant software involved.

We have sat in that chair and run this exact workflow with owners across Ohio, and the win is not some futuristic system. The win is that the boring first draft of the schedule, the prep sheet and the supplier note gets written by something that does not get tired, while you keep the judgment calls.

What AI actually does here

A tool like ChatGPT or Claude is a general assistant. It does not know your restaurant, your staff or your suppliers until you tell it. What it is good at is turning a messy description into a tidy draft. You give it the constraints, the roles, the opening hours and the quirks, and it returns a schedule grid you can correct. You give it your par levels and it hands back a prep sheet. You give it the situation and it writes the supplier email in a normal, professional tone.

The word prompt just means the instruction you type. A good prompt has three parts: the role, the constraints and the output format. "You are helping a restaurant manager. Build a weekly schedule for these roles and hours. Give it as a table." That is the whole skill. Everything below builds on that pattern.

Columbus is not a small market, and the competition for staff and customers is real. The city's 2026 budget overview lists a regional workforce of 1,182,200 people and a population growth rate of 1.4 percent, above the 1.0 percent national average, so the restaurants that run tighter weeks are the ones with room to breathe.

Who this is for and who it is not for

This is for owners and managers who personally build the schedule, write the prep lists and answer the supplier emails every week. You do the thinking, the tool does the typing. If you already have a scheduling platform that handles this, you may not need this workflow at all.

It is not for anyone hoping AI will run payroll, track labor law compliance or plug into a POS system. A general chat assistant does none of those things, and no source we found supports pretending otherwise. It drafts text. That is the honest limit, and the limit is fine, because drafting is the part eating your Sunday night.

Build the workflow in one sitting

Do this once, on a weeknight, in about thirty minutes. Save the three prompts somewhere you will find them, and next Sunday you paste, tweak and send.

The starter prompts, in full

Here is a fuller scheduling prompt you can copy. "You are helping a restaurant owner in Columbus. My staff: [names, roles, max hours, availability]. My hours: [opening hours per day]. Build next week's schedule as a table, one row per day, one column per person. Do not schedule anyone outside their availability. Flag any gap where I have no coverage." The flag request matters. It turns the draft into a checklist of the real decisions only you can make.

For prep: "Create a weekly prep sheet for a restaurant kitchen. Stations: [list]. Par levels: [list]. Expected covers per day: [numbers]. Output one section per station, each item with a quantity and a prep note." Paste your actual pars, because the tool has no idea what your walk-in looks like.

For suppliers: "Draft an email to my produce supplier. Situation: [describe the issue]. Keep it polite and under 120 words, ask one clear question, and sign off as [your name], [your role]." Supplier correspondence is where the tone improvement is most obvious. Short, clear emails get faster answers.

What not to hand over

Keep three categories out of the chat window. First, private employee details. Do not paste full names with phone numbers, addresses, wages or anything a staff member would not want sitting in a server elsewhere. Use first names and roles instead. Second, anything with legal weight: hiring decisions, discipline, terminations and payroll are yours and possibly your lawyer's, not a chatbot's. Third, anything requiring taste. The tool cannot taste the soup, judge a new hire's fit or read the room on a busy Friday. It drafts. You decide.

The safe habit is simple. Treat everything you paste into a chat as if a stranger could read it. If that thought bothers you, cut it from the prompt.

Where we come in

If you want a hand building this, that is our work. At Ohio AI Lab we run an AI Concierge, a real person who learns your business, holds two 1-hour working calls a month on tasks like these, and answers voice messages and email between calls with no limit on questions. Everything we build lives in your own accounts and is yours to keep. If you would rather train your kitchen manager or front-of-house lead on this workflow, our AI Training for Business runs 90-minute hands-on sessions built on your approved real tasks, and people leave with take-home routines and safe-use ground rules. The first step either way is a free 25-minute fit call on Zoom that looks at two tasks that repeat in your week and picks a first result worth having.

Questions worth asking before you start

Start with one prompt, the scheduling one, this Sunday. If the draft saves you even twenty minutes and you trust the output, add the other two next week. That is the whole build, and it costs you one evening.