Capture every event in one structured client brief, paste it into a chatbot with your house rules, and ask for a timeline worked back from service, an equipment list, a staffing plan and a load-in plan. A manager checks the pack line by line before anyone orders food. The brief is the input; the review is the safety net.
Printable catering checklists list the tasks every event has and miss the ones that make this event different, which is where things go wrong. A run sheet built from the brief starts from the facts of one event: 80 guests, a 6:30pm buffet, a hall with a prep kitchen and a 30-minute window to unload. The event below is invented for illustration, and every ratio and capacity in it is an assumption to replace with your own.
One brief in, six documents out
A well-formed brief can produce the whole pack, so you gather the facts once. The six documents are the event-day run sheet, a prep schedule for the days before, an equipment list, a staffing plan, a load-in and load-out plan, and a dietary and allergen plan. They depend on each other. The equipment list follows from the menu, the staffing plan from the guest count and service style, and the run sheet from the access times. Changing one fact in the brief, such as the unloading window, should change several documents, which is why a chatbot that rebuilds the pack from the brief beats six separately edited files.
The weak point is the brief, not the chatbot. If the brief does not say how long you have to unload, the chatbot will invent a number that sounds reasonable. So the first job is a template that forces the facts out of the client and the venue.
The brief template: the fields that change the run sheet
EVENT BRIEF (one per event)
Client (first name only):
Event type and date:
Venue and kitchen: ovens (how many, how many trays each), hobs, fridges, sinks, hot water, dishwasher, sockets and any power limit
Access: earliest arrival, unloading window, parking, route, steps or lift, time we must vacate
Guests: total, children, and counts by dietary group (counts only; names stay on the client's own list)
Service: guest arrival, food service, dessert, drinks, last food, finish
Menu: each dish, portions, where it is cooked, how it is held and served
Hire and extras: crockery, linen, chafing dishes, urns, who supplies each and when it arrives
Staffing: what the venue or client provides, what we provide
Contacts on the day: role and first name only
Sign-offs still needed: final numbers date, menu, dietary list
Questions we could not answer:Each group is there because it changes something downstream.
| Field group | What it changes | Fault it prevents |
|---|---|---|
| Venue kitchen and power | What is cooked where, and how food is held and heated | A menu that needs more oven space or power than the hall has |
| Access and unloading window | Arrival time, load order, number of carriers | Arriving late into a short window and blocking the only bay |
| Dietary counts | Separate prep, plates, labels and the order of service | A severe allergy handled as an afterthought |
| Service times and vacate time | The whole timeline, worked backwards, and the length of clear-down | Food late to the table, or a 90-minute clear-down squeezed into 40 |
| Hire and extras | Equipment list and who checks deliveries | Hired items nobody counted before the guests arrived |
| Sign-offs and questions | What is still open | Building a pack on numbers the client has not confirmed |
If clients send enquiries as long emails, use a chatbot to extract the fields into the template and to list what is missing, then send the questions back. The event planner intake form tutorial covers collecting these details before a quote, so the brief arrives half-complete.
A filled-in brief: 80 guests, a hall with a prep kitchen
This is the worked event, with the facts that drive the plan.
| Field | Entry (illustrative) |
|---|---|
| Event and date | 60th birthday party on a Saturday |
| Venue and kitchen | Hired community hall, ground floor. Prep kitchen with two domestic ovens, a four-ring hob, one large fridge, no dishwasher. Two double sockets. |
| Access | Kitchen free from 1:00pm. Rear door, one parking bay, 30-minute unloading window. Hall must be vacated by 11:30pm. |
| Guests | 80 in total, including 4 children under 10 (half portions). 14 vegetarian, 6 gluten-free (2 of them coeliac), 2 with a severe nut allergy. |
| Service | Guests arrive 6:00pm. Buffet opens 6:30pm. Client's cake cut and served at 8:00pm. Tea and coffee from 8:15pm. Last food out 9:00pm. |
| Menu | Chicken in mustard cream sauce (cooked at base, chilled, reheated on the hob). Vegetable lasagne and roast potatoes (oven). Green beans (hob). Three salads and bread rolls. |
| Hire and extras | Client's hire company delivers 90 plates and 90 sets of cutlery at noon. We supply chafing dishes, two urns and serving gear. |
| Staffing | Nobody provided by the venue. |
Notice what the brief does not contain: a guest name, a phone number or a street address. Dietary needs are counts, and the matching of plates to people stays with the client's list. That keeps health information out of the chat and still gives the chatbot everything it needs for the plan.
House rules the chatbot cannot guess
Keep a short file of the rules you already run by, and put it in a ChatGPT Project or a Claude Project so every event pack starts from the same file. These are the assumed house rules for the example.
- Buffet: one front-of-house server per 25 guests, minimum two. Plated service: one per 12.
- Event lead on site for any event over 50 guests.
- Two cooks minimum when food is reheated or finished on site.
- Add 15 minutes to every travel estimate.
- Assume two trays per domestic oven unless the venue tells us otherwise.
- A 30-minute break for any shift over six hours, placed in a quiet window.
- Two spare serving utensils per serving line, and spare chafing fuel.
- One urn per 50 guests for hot drinks.
- Food-safety times and temperatures come from our own food-safety plan, never from the chatbot.
The last rule matters. A chatbot will state holding times and temperatures with confidence, and they differ with the rules where you work and with your own plan. Tell it to write "per our food-safety plan" and leave the numbers to the people trained in them.
The prompt that builds the pack
You are a catering operations planner. Build an event pack from the brief below, using my house rules.
Produce, in this order:
1. Event-day run sheet: a table with time, task, who and a check, worked BACKWARDS from the start of food service. Include travel, the unloading window, set-up, hot and cold holding, service, clear-down, load-out and the time we must vacate.
2. Prep schedule for the three days before.
3. Equipment list grouped by cooking, hot and cold holding, transport, service, cleaning and waste, safety and admin. Give a quantity and the reason for each quantity.
4. Staffing plan using my ratios, with arrival and finish times and a break for any shift over six hours.
5. Load-in and load-out plan: load order (last in, first out), who carries what, route.
6. Dietary and allergen plan using only the counts in the brief.
Rules: use only facts in the brief. If a fact is missing, list it under QUESTIONS at the end instead of assuming it. Show your sums for quantities and times. Do not state food-safety temperatures or times; write "per our food-safety plan". Flag any step where venue equipment capacity could be a problem.
HOUSE RULES: [paste]
BRIEF: [paste]Three clauses carry most of the weight. "Worked backwards" makes the chatbot start from service time. "List missing facts under QUESTIONS" stops it filling gaps with plausible guesses. "Flag any step where venue equipment capacity could be a problem" turns a hidden weakness into an item on the page.
The first run sheet, and the five faults a manager caught
Here is the main stretch of a first-draft event-day run sheet, trimmed (illustrative output).
| Time | Task | Who |
|---|---|---|
| 12:30 | Load vehicle at base kitchen | Cook 1, runner |
| 13:00 | Arrive at hall and unload | Event lead, cook 1, runner |
| 13:30 | Move vehicle off the bay; set up kitchen | Runner, cook 1 |
| 17:15 | Put lasagne and roast potatoes into ovens 1 and 2 | Cook 1, cook 2 |
| 18:00 | Guests arrive | Event lead |
| 18:30 | Buffet opens | All servers |
| 21:00 | Clear-down | All staff |
| 22:00 | Vehicle loaded and gone | Runner, cook 1 |
It looks tidy and has five faults, each caught on a read-through.
- No travel time. Loading at 12:30 and arriving at 13:00 ignores a 40-minute drive plus the 15-minute buffer in the house rules. Leaving at 12:00 puts the van in the bay at 12:55 for a 1:00pm start.
- Oven capacity. All the oven food goes in at one time, which the ovens cannot hold. The next section shows the sums.
- No breaks. The event lead, cook 1 and the runner all work more than six hours, and the sheet gives none of them a break.
- A 60-minute clear-down for a 90-minute job. Food is out until 9:00pm and the hall must be empty by 11:30pm, so there is room for a realistic clear-down, yet the draft allowed an hour and had the vehicle gone by 10:00pm, which would send staff home with the hall half cleared. Nothing in the brief or the house rules said how long clear-down takes, so the chatbot guessed short. Put your own figure in the house-rules file and it stops guessing.
- No allergen step. There is no moment when the nut-free plates are prepared, labelled and handed over, because the brief says "2 with severe nut allergy" and the first draft never turned that into a task.
Feeding the faults back ("fix these five and show what changed") produces a better second draft in seconds. Do not skip the human read of the first one: the faults are easy for a manager who knows the venue and invisible to a chatbot that does not.
Oven maths: seven trays, four slots
The run sheet had everything entering the ovens together. Test it with a sum. Lasagne for 24 portions needs two full-size pans. Roast potatoes for 80 need five trays at about 16 portions a tray. That is seven trays. Under the house rule of two trays per oven, the two ovens hold four trays at a time, so seven trays need two cycles. At roughly 65 minutes a cycle, including a short turnaround, that is about 130 minutes of oven time, and the earlier the first cycle starts, the more slack there is before the buffet opens.
The corrected plan puts lasagne (two pans) and two trays of potatoes into the ovens at 15:30, and the last three trays of potatoes at 16:40, which finishes at about 17:45. All the oven food moves into hot-holding by then, and the chicken reheats on the hob from 17:15. The buffet opens at 18:30 and the hot-holding step follows your own food-safety plan. If the oven maths had come out at four cycles instead of two, you would change the menu, hire a portable oven or cook more at the base kitchen, and you would tell the client before the event, not on the day. The tutorial on scaling recipes and shopping lists covers the quantities behind each tray.
Equipment list: from the menu, then the invisible items
Ask the chatbot to derive equipment from the menu and give the reason for each quantity, then add what no menu mentions. Anything it cannot explain with a sum gets questioned.
| Group | Item and quantity | Basis |
|---|---|---|
| Cooking | Full-size roasting trays x9 | 7 needed (2 lasagne, 5 potato) plus 2 spare |
| Cooking | Large pots x4 | Four hob rings: chicken x2, beans, one spare |
| Hot-holding | Insulated food carriers x3 | Hold the seven oven trays between oven and buffet, up to three per carrier |
| Hot-holding | Chafing dishes x8, plus fuel | Two serving lines of four hot items each |
| Service | Serving spoons and tongs x20 | Two lines x 8 items (4 hot, 3 salads, bread), plus 2 spare per line |
| Service | Salad bowls x6, bread baskets x4 | Three salads on two lines; bread one basket per table end |
| Hot drinks | Urns x2, cups and saucers x90 | One urn per 50 guests; cups per guest plus spares |
| Safety | Probe thermometers x3, sanitiser wipes, labels | One per work area, plus allergen cards |
| Cleaning | Hand-wash soap, paper towels, bin bags, cloths | Not listed in any menu, so easy to forget |
| Admin | Printed run sheets, role sheets, extension lead, first-aid kit | Every person gets a page; one extension lead per work zone if the power rules allow |
The invisible items sit in the last three rows. They are the ones that cause a 5pm run to the shop, so the manager adds them from experience and then stores them in the house-rules file, so the next pack includes them without being told.
Staffing: ratios, shifts and what they cost
Four servers come from the buffet ratio: 80 guests divided by 25 is 3.2, so round up to four. The house rule also requires an event lead, and two cooks because food is reheated on site, plus one runner for dishes and clear-down. That is eight people. The chatbot builds the shifts and the manager checks the hours.
| Role | Number | On site | Rostered hours each | Total hours |
|---|---|---|---|---|
| Event lead | 1 | 12:00 to 23:00 | 11.0 | 11.0 |
| Cook 1 | 1 | 11:45 to 21:30 | 9.75 | 9.75 |
| Cook 2 | 1 | 15:00 to 21:30 | 6.5 | 6.5 |
| Runner and dishes | 1 | 11:45 to 23:00 | 11.25 | 11.25 |
| Servers | 4 | 16:30 to 22:30 | 6.0 | 24.0 |
| Total | 8 | 62.5 |
At an assumed loaded cost of $22 an hour that is $1,375, or $17.19 a guest, which is the number to take back to your price. The cost-per-person pricing sheet tutorial shows where it goes. Check three things yourself. Every shift over six hours has a 30-minute break in a quiet window (the first draft gave none), the servers are present before the first guest arrives, and nobody finishes before their last task. The rules on breaks and working hours differ by place, so use your own, not the chatbot's.
Load-in and load-out in a thirty-minute window
The unloading window is a constraint, not a suggestion: the vehicle has to leave the only bay by 1:30pm. Ask the chatbot to plan the load as last in, first out, with three people carrying and two trolleys, and to give each crate a number that appears on the load card.
| Unload order | Crates and items | Goes to |
|---|---|---|
| 1 | Cold boxes: chilled chicken, salads, dairy | Fridge |
| 2 | Roasting trays, pots, insulated carriers | Hob and oven end of the kitchen |
| 3 | Serving gear, labels, allergen cards, thermometers | Buffet table |
| 4 | Chafing dishes and fuel | Buffet table, fuel locked away |
| 5 | Urns, cups, bread baskets | Drinks area |
| 6 | Cleaning kit, bin bags, first-aid kit | Back lobby |
The vehicle is packed in the reverse order, with cold boxes loaded last so they come off first. After unloading, one person moves the vehicle and the other two start the kitchen. Load-out works the same in reverse, with dirty, wet kit bagged and packed away from anything clean. The first draft of this plan put the cleaning kit by the van doors, which would have meant unloading it before the chicken. Reordering is exactly the job a chatbot does well, once it knows the rule.
Allergens: counts from the brief, facts from the labels
Let the chatbot draft the dietary plan from the counts: two severe nut allergies, six gluten-free guests (two of them coeliac), fourteen vegetarians and four children on half portions. A good first draft sets out what each group needs: a nut-free plate prepared first with clean equipment, covered, labelled and handed over by the event lead, a gluten-free plate for the six, a vegetarian main on the hot line and a note on each label. It can also build an allergen matrix from the recipes.
The matrix is where a chatbot is weakest, because it works from recipes and not from the products you actually bought. A realistic miss: the chicken sauce is thickened with cornflour, so the draft marks it gluten-free, but the stock cube in the supplier's version lists wheat. Only the label in front of you settles that. A person with food-safety training reads every product label for the real ingredients and signs the matrix off, and anything unclear is treated as containing the allergen. The tutorial on allergen icons and dietary labels before you print covers the labelling side. Nothing about allergens goes onto a printed label on the chatbot's word alone.
Twelve checks a manager signs before anyone orders food
This is the review step. Each check has a reason and a way to prove it, and the manager signs and dates the pack when all twelve pass.
- Counts match the brief. Why: every quantity depends on them. Verify: add the dietary groups and children against the total, and compare the portions per dish with 80.
- Times add up backwards. Why: errors compound. Verify: start at 18:30 and subtract each duration with a pencil, including travel and buffer.
- Venue limits are respected. Why: the venue's rules are fixed. Verify: unloading window, bay, route, vacate time and power each appear in a task.
- Oven and hob capacity. Why: the ovens are the usual bottleneck. Verify: trays needed, trays per cycle, cycles, and the finish time.
- Hot and cold holding. Why: food safety. Verify: for every dish, where it is at each hour, per your food-safety plan.
- Allergen plan. Why: the consequences are serious. Verify: product labels checked, separate prep, plate order and a named person.
- Equipment ties to the menu. Why: forgotten items cost time. Verify: each dish has a container, a heat source, a utensil and a label.
- Staff numbers, shifts and breaks. Why: tired staff make mistakes. Verify: ratios, arrival before guests, a break for shifts over six hours.
- Vehicle capacity and load order. Why: one trip fails if the van is full. Verify: crates counted against van space, with cold boxes last in.
- Contingencies. Why: something will slip. Verify: a written fallback for a late guest start, an oven failure and a late van.
- Client sign-offs. Why: the pack is built on numbers. Verify: final numbers, menu and dietary list are confirmed in writing.
- Version control. Why: old pages cause confusion on the day. Verify: one dated version, role sheets printed from it, earlier copies removed.
Once the manager signs, ask the chatbot to cut the master run sheet into one-page versions per role, with only that person's times, tasks, contacts, allergen rules and break. Print them. A page per person is easier to follow at 5pm than a six-page pack.
After the event: planned against actual
The pack improves only if what happened feeds back into the house rules. The next morning, record planned and actual times for the main steps and ask the chatbot for the cause and the rule change. Here is a short example (illustrative).
| Step | Planned | Actual | Cause | House-rule change |
|---|---|---|---|---|
| Unloading | 30 minutes | 38 minutes | Second trolley would not fit the rear door | Use one trolley and a third carrier for halls with narrow doors |
| Buffet opens | 18:30 | 18:41 | Client's speech ran long | Ask clients for a run order with a 15-minute buffer |
| Clear-down | 90 minutes | 75 minutes | Four servers helped the runner | Keep 90 minutes, but release servers at 75 |
Add each change to the house-rules file, and the next brief benefits without anyone having to remember it. The tutorial on event feedback and debriefs with AI goes further on the debrief itself. Over a season the house-rules file becomes the real asset: it holds your actual unloading speeds, your own oven capacities and your clients' habits, and no template from a website can.
Questions caterers ask about AI run sheets
Can ChatGPT build a run sheet from a client's email?
Yes, once the email has been turned into the brief fields. Paste the brief and your house rules, and tell the chatbot to list missing facts as questions instead of assuming them. An email alone leaves too much unstated, such as access times, power limits and the kitchen at the venue.
How much of an AI-built run sheet can I trust?
Treat it as a draft that is right about structure and often wrong about this venue. Counts, times, oven capacity and the load-in window all need a manager's check against the brief. The review step, not the chatbot, is what makes the sheet safe to use on the day.
Should guest names and allergies go into the prompt?
Use counts, such as six gluten-free and two with a severe nut allergy, and keep names on the client's own list. Allergy details about named people are health information. The run sheet needs the numbers and the handling rules, and the plates can be matched to people separately.
What should a staff member's one-page version contain?
Only their own arrival and finish times, their tasks in order, who to call, the allergen handling rules and their break slot. A single page per role is easier to follow on a busy afternoon than the full pack, and AI can extract each role from the master run sheet.
Further reads
- How Caterers Use AI to Quote Events and Plan Menus — The wider picture of how caterers use AI for quotes and menus.
- Event Planning Checklist: Build One Per Event Type With AI — Build a checklist per event type, not one generic list.
- DJ Event Checklist: Gear, Timeline and Backup Plan Built With AI — The same gear-and-timeline method for another event supplier.
- Catering Proposal Template: Packages, Prices and Terms With ChatGPT — Proposal wording that sets up the brief before the event.
- How Event Planners Use AI to Build Budgets and Run Sheets — How planners build budgets and run sheets with AI.
- AI in Food Service: A Four-Week Menu Cycle With Costs and Allergens — Costs and allergens across a repeating menu cycle.
- Catering Quote Software: Which Tools Turn Enquiries Into Quotes? — Time an enquiry-to-quote run with and without AI, then pick between caterer-specific and general quote tools using dated list prices.
- Catering Software for Small Business: Free Options and AI Extras — A free quote, order and invoice stack for a small caterer, then what a free AI assistant adds to each step and where it stops.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: every figure in the worked event, including oven capacity, staff ratios and wage rates, is an assumption for illustration; replace it with your own house rules.