Use AI for the paperwork-shaped jobs in four weeks: week one for name checks and a break-even sum, week two for supplier emails and quote comparison, week three for the menu or range and a draft cash-flow forecast, week four for the Google profile and opening-week posts. AI drafts each item. A person checks every number, name and supplier term.
Here "first month" means the four weeks before the doors open, with the lease signed or about to be: one owner, perhaps one helper, and a budget measured in thousands. A cafe and a shop need different menus and ranges but the same calendar, and the order matters, because the supplier choice, the prices and the size of the first stock order all depend on the break-even sum from week one. Tool choices and their prices are in starting a business using AI: tools and costs for a small shop.
Week 1: a name that survives checking, and the break-even sum
A name is the first job AI does well and the first one it can do harm on. It is fast at generating options against criteria. It cannot reliably search a trade mark register, and it will happily call a name "available" when it has no way to know. So use it for the long list and use people for the checks.
I'm opening a [14-seat cafe / gift shop] on a busy residential high street.
Give me 30 name ideas. Criteria: easy to spell after hearing it once; one
or two short words; no puns that need explaining; readable on a sign from
across the road; not a person's name; nothing close to these well-known
chains: [list]. For each, say in eight words why it fits. Then flag any that
sound like an existing brand or that many businesses probably use.Expect a list that mixes usable ideas with claims the assistant cannot back. An illustrative slice of a reply:
4. Corner Cup: short and literal, easy to read from across the road.
9. Steam & Crumb: pairs coffee with baking; the ampersand may trip people who only hear it.
15. The Daily Grind: classic coffee wordplay that customers recognise.
Flags: none of these appear to be registered trade marks.
Three edits before anything reaches the shortlist. Keep the ampersand warning, because spelling traps are something the assistant spots well. Strike "The Daily Grind": it runs to three words, and customers recognise it because many cafes already use it, which is exactly what the prompt asked the assistant to flag. And delete the flag line, because the assistant has not searched any register and has no way to know.
Take the best five and run each through a short set of checks. The table shows what a pass and a fail look like:
| Check | How | A fail looks like |
|---|---|---|
| Say it, spell it | Say the name to three people on the phone and ask them to write it down | Two of the three write a different spelling |
| Trade mark | Search your market's trade mark register for food, drink or retail marks | An identical or near-identical mark in your field. Ask an adviser if it is close |
| Domain | Check the .com or your market's usual domain, and the renewal price, not only the first-year price | Taken, or available only with a hyphen or an odd spelling |
| Social handles | Look for the same handle on the platforms you will use | Taken by an inactive or unrelated account |
| Nearby clutter | Search the name plus "cafe" or "shop" in a map app | A similar name two streets away |
| Sign test | Type it at sign scale in the font you want | Letter pairs that blur, such as "rn" reading as "m" |
You can ask the assistant whether a name means anything awkward in other languages your customers speak, then confirm with a native speaker. Treat its answer as a lead, not proof. Generating and checking AI brand names before you buy goes deeper on the trade mark side.
The other week-one job is the break-even sum, because every later decision depends on it. Take a 14-seat cafe open six days a week, about 26 trading days a month, with an average spend of $9.50. These figures are an illustration, and you replace each with a real quote as it arrives.
| Monthly fixed cost | Illustrative amount |
|---|---|
| Rent | $3,200 |
| Wages (not including the owner's own pay) | $7,000 |
| Utilities | $450 |
| Insurance | $160 |
| Software, phone and internet | $180 |
| Supplies, repairs, marketing and bookkeeper | $670 |
| Total | $11,660 |
Variable costs come to about 33.5% of sales: roughly 30% for ingredients, packaging and waste, and 3.5% for payment fees, allowing for some cash sales. That leaves a contribution margin of 66.5 cents in every dollar. Break-even sales are fixed costs divided by that margin: $11,660 divided by 0.665 is $17,534 a month, which is $674 a day, or 71 customers a day at $9.50. That one figure shapes the rest of the plan. Ask the assistant to build the sheet, with the formulas typed in cells, and check the result once with a calculator. A chat assistant that adds up a long column in its reply can get it wrong, whereas a formula in a cell recalculates the same way every time.
Done when you have a shortlist that passed the checks, an application started for any name or business registration you need, and one break-even figure you trust.
Week 2: supplier emails that get a price, and a sheet to compare them
Suppliers answer clear emails from people who sound as though they will order. The weak point of an AI draft is that it sounds like an established business that already buys in bulk, which suppliers see through. Give it your real volumes and your real questions. This prompt works for a roaster, a bakery, a wholesaler or a maker:
Write a 90-word email from a new [cafe / gift shop] opening on [date] to
[supplier type], asking for a trade account. Plain, polite, no hype.
Ask for: price list, minimum order, delivery days, payment terms for a
new customer, how long product stays at its best, and whether samples
are available. Say our expected first-month volume is [about 25 kg of
coffee / a first order of about $1,200]. Use [your name] and [phone].
Do not claim we are established or promise future volumes.A first draft often includes lines such as "we plan to order 500 units weekly" or "we're excited to explore a long-term partnership". The first is an invented volume that could be held against you, and the second is filler. Delete both. Replace them with what you actually expect to buy in month one. For a gift shop that stocks local makers, add one more line that costs nothing: ask whether they offer sale or return for a first order, which lowers your opening stock risk.
When quotes come in, put them in one sheet. Here is an illustrative comparison for coffee, using month-one use of about 25 kg (45 customers a day, 1.2 coffees each, 18 g a cup, 26 days):
| Supplier | Price per kg | Minimum order | Delivery | Terms | Month-one cost |
|---|---|---|---|---|---|
| A | $24.00 | 10 kg | Weekly | First order prepaid, then 14 days | $607 |
| B | $21.50 | 25 kg | Monthly | 30 days | $544 |
| C | $26.00 | 5 kg | Weekly | Prepaid | $658 |
The gap between cheapest and dearest is $114 a month, about 1% of the month-one sales in the cash flow below. Supplier B saves $63 against A, but delivers once a month, so a month's coffee sits on your shelf, and its 25 kg minimum is your entire month-one use. That is a reason to ask how long the beans stay at their best before choosing on price. AI can paste quotes into this layout and flag mismatched units, which is useful; comparing supplier quotes side by side with AI shows the method. Three things it cannot judge for you: taste (do a tasting), reliability (ask for two references from other cafes or shops) and whether the credit terms suit your cash flow.
Done when each key product has two or three quotes in one sheet, samples are ordered, and you have written down why you favour one supplier for each product.
Week 3: price the menu or the range from real costs
For a cafe, start with the drinks, because a handful of lines make most of the sales. Cost one cappuccino. Coffee at 18 g and $24 a kilo is 43 cents. Milk at 180 ml and $1.10 a litre is about 20 cents. Cup, lid and sleeve are 16 cents, and sugar and a stirrer another 2 cents. That is about 81 cents. At $4.20 the ingredients are about 19% of the price. Then check the payment fee, because a flat per-sale fee hurts small tickets. A payment provider that charges 2.6% plus 15 cents per card sale (the in-person rate Square lists on its Free plan; check your own provider) takes about 26 cents from a $4.20 coffee, or 6.2%, but only about 40 cents from a $9.50 order, or 4.2%. Adding a pastry to the order spreads the fixed fee over a bigger ticket, so it is worth more than the pastry's own margin.
Give the assistant your ingredient quantities and quote prices and ask for a cost-per-item table with a column for percentage of price. Then check three rows by hand. Coffee shop pricing: cost per cup and margins with AI walks through the whole drinks list. For menu wording, ask for two sentences per item in your voice, and delete any claim you cannot back, such as "locally sourced" when you have not checked where it comes from.
A shop plans a range, not a menu. Suppose a gift shop has $12,000 to spend on opening stock, measured at cost. The grid below splits it across six categories with a planned markup on each, where markup is retail price divided by cost. The figures are illustrative.
| Category | Share | Spend at cost | Markup | Retail value |
|---|---|---|---|---|
| Cards, wrap and stationery | 14% | $1,680 | 2.6 | $4,368 |
| Candles and home scent | 20% | $2,400 | 2.2 | $5,280 |
| Kids and games | 16% | $1,920 | 2.1 | $4,032 |
| Kitchen and tableware | 18% | $2,160 | 2.0 | $4,320 |
| Jewellery and accessories | 12% | $1,440 | 2.5 | $3,600 |
| Seasonal and gifting sets | 20% | $2,400 | 2.2 | $5,280 |
| Total | 100% | $12,000 | 2.24 blended | $26,880 |
The blended markup of 2.24 means a gross margin of about 55% before discounts, damage and theft, so plan for a little less. Now test the stock against the shop's break-even. With fixed costs of $5,900 a month and card fees of about 3%, the contribution margin is about 52 cents in the dollar, so break-even sales are about $11,270 a month. At an average basket of $32, that is roughly 350 sales a month, or 13 or 14 a day. The cost of the goods sold at break-even is about $5,030, so $12,000 of opening stock is about 2.4 months of cover. You may be comfortable with that, or you may want to buy less and reorder. Either way, the sum shows you the question before the money is spent. Bundles and follow-up offers come later, once you are trading.
Done when every product has a cost, a price and a margin you checked, and the opening order is sized against the break-even sum.
Also in week 3: the cash-flow draft a person must check
Ask the assistant for a first draft of a monthly cash-flow forecast for the first six months. Give it the fixed costs, the variable cost percentages, the break-even sum and your pre-opening spending. Its job is to build the structure with formulas, not to guess. Here is the mistake to expect. Feed it "break-even is 71 customers a day" and a careless draft assumes 71 customers a day from day one, which makes every month cash-neutral and the plan look safe. The checked version ramps up:
| Month | Customers a day | Sales | Net after all costs | Cumulative |
|---|---|---|---|---|
| 1 | 45 | $11,115 | -$4,269 | -$4,269 |
| 2 | 58 | $14,326 | -$2,133 | -$6,402 |
| 3 | 70 | $17,290 | -$162 | -$6,564 |
| 4 | 80 | $19,760 | $1,480 | -$5,083 |
The first draft shows roughly zero loss each month. The ramp shows the cafe needs about $6,564 to cover its early losses, before counting pre-opening spending. Say that spending is $22,000 for fit-out, equipment, deposits and first stock. Add a buffer of one month's fixed costs, $11,660, and the cash the owner needs before the cafe pays its way is about $40,000. A cash-flow tab that said "safe from month one" would have sent the owner to the bank for about $22,000 and left roughly $18,000 to chance.
A person must check the draft. One hour with a bookkeeper or accountant is the best use of money in the month. Take this checklist:
- Opening balance matches the real bank balance, not a round number.
- The ramp is tested against your own footfall count: stand outside for two hours at three different times and count people passing, then compare with the customers a day you assumed.
- Wages start before opening for training week, and include employer costs such as payroll taxes and pension contributions. Ask your accountant what applies to you.
- Rent shows the deposit, any rent-free period and when full rent starts.
- Stock is bought before the first sale, so it hits cash earlier than cost of sales.
- Supplier credit terms move cash out on the right date, not the invoice date.
- Tax you collect on sales sits on its own line, because it is not your income.
- Totals are formulas, not numbers typed in by an assistant. Change one input and see whether the totals move.
For the method behind a forecast like this, building a 13-week cash flow forecast with AI help goes through it step by step.
Weeks 3 and 4: Google profile first, then a week of launch posts
Claim your Google Business Profile as early as you can, because verification takes time. Google chooses the verification method, and you may see only some of the options: phone or text, email, a live video call, a postcard or a video recording. According to Google's help page, its review of your verification can take up to five business days, and most postcard codes arrive within 14 days. A live video call is only available during your business hours and asks you to show your location and that you run the business, for example outside signs, equipment or the till. So aim for the week the sign and equipment go in, which in most plans is week three, and not while the unit is still bare. Creating a Google Business Profile: verify, categories, photos and AI text covers the rest of the setup.
One rule matters in opening week. Google's review policy bars offering incentives for reviews and selectively asking only happy customers. A "free coffee for a five-star review" offer is the classic breach. Ask everyone, offer nothing in return, and make it easy: a card at the till with a link or QR code.
Plan seven posts and give each one a job. Ask the assistant for drafts using real facts you supply: dates, hours, address, three real products.
| Day | Post | Check before it goes live |
|---|---|---|
| 7 days before | "We open on [date]" with opening hours | Hours match the door sign and the Google profile |
| 3 days before | Behind the scenes: the machine, the shelves, the team | Everyone shown has agreed |
| 1 day before | Menu or range preview with your own photo | Prices are final |
| Opening day | "We're open" with a simple opening offer | The offer is real and has an end date |
| 2 days after | Thank-you to the first customers, no names without consent | No customer photos without permission |
| 4 days after | Spotlight on one product or supplier | Claims about the supplier are true |
| 7 days after | First-week recap and a plain ask for feedback | No incentive attached to reviews |
Expect the usual AI habits in the drafts: exclamation marks, "grand opening" hype, made-up offers such as "50% off all week" and long hashtag lists. Instagram has capped hashtags at five per post since December 2025, so trim the list. Cut anything you cannot honour. If you use AI images, never show food or products you do not sell. The cafe photo test in the further reading shows what looks honest. A soft opening a few days earlier for friends and neighbours is the final test: time the orders, watch the queue, try the card machine, and write down every fix before the real doors open.
Done when the profile is verified or in progress, seven posts are drafted and approved, and the soft opening has produced a fix list.
The four weeks on one page, with hours and who does what
| Week | AI does | A person does | Owner hours |
|---|---|---|---|
| 1 | Long list of names; break-even sheet with formulas | Name checks; registration forms; real fixed-cost quotes | 8 |
| 2 | Supplier emails; quote comparison layout | Tastings; references; terms decisions | 8 |
| 3 | Cost-per-item table; range grid; cash-flow draft | Check three rows by hand; footfall count; bookkeeper review | 9 |
| 4 | Seven post drafts; captions; alt text | Profile verification; photos; soft opening; fix list | 8 |
That is about 33 hours of owner time, most of it checking and deciding. The AI side costs between nothing and about $20 for the month: free assistants cover the drafting, and a $20-a-month paid plan is a list-price option if you hit usage limits.
What I would leave until month two
New owners tend to want the clever part first. These can wait until you are trading and have real data:
- A website chatbot. You do not yet know what customers ask. Once you do, a chatbot FAQ and offers for your quiet hours both become easier.
- Loyalty schemes and automations. They need customers and a till that records them.
- Paid ads. Wait until you know your average spend and your quiet hours.
- AI phone answering. Opening hours and a phone number on your Google profile come first.
The first month is for decisions that are expensive to reverse: the name, the suppliers, the stock and the cash you need. AI saves hours on the drafting of each. Only you, your bookkeeper and your own footfall count can say whether the numbers are right.
New shop and cafe owners ask about AI in month one
Which AI tool should a new owner start with?
A free chat assistant is enough for week one: the free tiers of ChatGPT, Claude and Gemini all draft emails, posts and spreadsheet formulas. Move to a $20-a-month plan only if you hit usage limits. Spend your first spare money on an hour of a bookkeeper's time to check the cash flow, not on more software.
Should AI write my business plan?
It can draft the structure and tidy the wording, but the numbers must come from your quotes, your lease and your own footfall count. A plan padded with AI-estimated figures gives a lender or landlord reasons to doubt the rest. Use AI to explain and format; use your own data to fill in.
Can I use AI-generated food photos for opening week?
Avoid pictures of food you do not serve, because customers compare the photo with the plate. Take your own photos on a phone in daylight. AI is useful for captions, crop suggestions and alt text, and for planning which shots you still need before opening day.
When should I deal with food-business registration?
In week one. Some authorities require registration before you trade, and inspections or paperwork can take longer than owners expect. This plan cannot tell you what applies to your premises, so ask the authority that covers them and write the dates into your calendar.
How many opening-week posts do I need?
About seven over two weeks, with a plain 'we're open' on the first day. The quality of the first three matters more than volume. Schedule only posts you have checked, because a scheduled post with the wrong opening hours is the most common early mistake.
Further reads
- Coffee Cart Business Plan: Costs, Pitch and Menu Built With ChatGPT — A smaller-scale plan for a coffee cart.
- Restaurant SOP Template: Opening and Closing Checklists With AI — Opening and closing checklists for your soft launch.
- AI-Generated Marketing Photos: A Cafe Test of What Looks Honest — What looks honest in AI-assisted cafe photos.
- How AI Can Help in Sales: A Gift Shop's Bundle and Follow-Up Plan — Bundles and follow-ups for a gift shop.
- How Cafés Use AI to Fill Quiet Hours With Targeted Offers — Fill quiet hours once you are trading.
- Best AI for Shop Design: Floor Plan and Display Brief for Your Fitter — A shop owner's guide to AI for a refresh: measure, draft a layout, mock up windows and signs, then hand the fitter a brief they can price and check.
- Board Game Café Business Plan: Table Pricing, Menu and Bookings With AI — How a board game café can price table time, cost a game-friendly menu, design booking slots and plan year one, with AI drafting and a spreadsheet checking.
- Incorporating AI Into the Workplace: A First-Month Plan for Small Teams — One task, one tool, one owner, four weekly reviews and a one-page use policy: how a small team can start using AI at work without the chaos.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Google Business Profile Help, 'Verify your business on Google' (verification methods and timings); Google's review policy; Instagram's five-hashtag limit (December 2025); Square's published in-person card rate on its Free plan (2.6% plus 15 cents). All business figures are illustrations.