Use AI at five points: testing the idea, choosing a name, building the store, finding first customers and setting up admin. A one-person shop can run its tools for about $40 a month lean, $96 standard or $156 comfortable, before payment fees, stock and rent. Start free, and pay only when a limit stops you.
Tool costs are a small part of starting a shop, and they should stay small, because no tool in the stack sells an item for you. The plan below follows one illustrative one-person stationery and craft shop that sells online and from a weekend stall, and it puts a price on every line so you can see how many orders each budget needs.
The monthly tool budget at three levels
As of October 2026, these are USD list prices from the vendors' own pages, on yearly billing unless stated. Vendors change them, so confirm each one before you pay. Card fees are a percentage of sales rather than a monthly line, so they sit outside the totals.
| Line | Lean | Standard | Comfortable | What the line buys |
|---|---|---|---|---|
| AI assistant | $0 | $20 | $20 | Free tiers first; then one paid plan (Claude Pro is $17 on annual billing or $20 monthly; the other main paid tiers list around $20) |
| Store | $39 | $29 | $29 | Shopify Basic: monthly billing in the lean case so you can stop, yearly billing once the shop works |
| Domain | $0.92 | $0.92 | $0.92 | A .com at $11.08 a year, same on renewal, with free privacy, at one registrar |
| Business email | $0 | $7 | $14 | A free address; Google Workspace Business Starter $7 or Standard $14 |
| Design | $0 | $12 | $12 | Canva Free; Canva Pro at $144 a year |
| Accounting | $0 | $27 | $32 | A spreadsheet; Xero Early $27, plus a $5 auto-reconcile add-on in the comfortable case |
| Email list | $0 | $0 | $13 | Mailchimp Free (250 contacts, 500 sends a month); Essentials from $13 at 500 contacts |
| Social scheduling | $0 | $0 | $15 | Buffer Free; Essentials at $5 a channel, three channels |
| Automation | $0 | $0 | $19.99 | Zapier Free; Professional at $19.99 a month on yearly billing |
| In-person payments | $0 | $0 | $0 | Square's Free plan, with a fee on each sale |
| Monthly total | $39.92 | $95.92 | $155.91 | |
| Orders to cover it | 4 | 8 | 13 | Using the sample order below |
The sample order is a $28.00 basket of goods plus $4.00 postage charged to the customer. The goods cost $12.60 (45%), packaging is $1.20, the postage you pay is $4.00, and the card fee at 2.9% plus 30 cents on $32.00 is $1.23. That leaves $12.97 an order. The lean stack of $39.92 needs 3.1 of them, so 4. The standard stack of $95.92 needs 7.4, so 8. The comfortable stack of $155.91 needs 12.0, and twelve orders fall 27 cents short, so 13.
Three things to read from the table. First, the lean stack is not a sacrifice: the free tiers cover everything a shop needs in its first weeks except a store and a domain. Second, the biggest single lines are the store and the assistant, so those are the two to justify. Third, these sums leave out stock, rent if you take a unit, insurance and your own pay, all of which dwarf the software. The tools are only worth buying early if each one saves you an hour you would otherwise spend.
Some introductory pricing will flatter the early months. Shopify's page offers a new store three days free, then $1 a month for three months. Google Workspace's page shows half price for new customers' first 12 months on the first 20 users. Both are worth taking, and neither belongs in the plan, because the list price is what you will pay once the offer ends. Card rates and the payment provider also depend on where you trade, so check the rate your own account is given. For what AI costs a shop over a whole first year, including paid add-ons for customer service, what AI costs a small retail shop in its first year goes into more detail than this table.
Week 1: Test the idea on real people before you pay for anything
The first job for an assistant is to attack the idea, not approve it. Asked whether a shop is a good business, a chatbot will say yes in well-formed paragraphs. Ask instead for the assumptions that would have to be true, and for a cheap test of each. This costs nothing on a free tier.
I'm thinking of opening a small stationery and craft-supplies shop, selling online and from a weekend stall. Don't tell me whether it's a good idea. List the ten assumptions that would have to be true for it to work (price people will pay, how they find me, repeat purchases, supplier terms, postage). For each, give one cheap test I can run in two weeks with a result I can count, and mark any assumption where you would only be guessing a number.An illustrative extract of the reply, with the pass line added by the owner before testing:
| Assumption | Cheap test | Pass line |
|---|---|---|
| People will pay $16 for a notebook that costs $7.20 delivered | Show three notebooks at a weekend stall at $16 | 5 sales from 30 people who stop |
| A supplier will sell in small quantities | Ask five suppliers for terms | Two accept a first order under $500 |
| Postage fits the basket | Weigh ten typical baskets and get carrier quotes | Postage at most 15% of the basket |
| Buyers come back | Ask ten buyers what they would buy next | Four can name a product |
| Instagram brings visitors | Post eight times in two weeks | 40 visits to the shop link |
Two fixes before you use it. Delete any market statistic the assistant adds, such as a claim that the category is growing steadily, unless you can open the source. An invented market size can send you building for demand that is not there. And treat the pass line as the owner's job: fix it before the test, so the result cannot be argued into a pass afterwards. For a deeper treatment of testing demand, see AI tools for startup ideation and validation.
Then put real numbers on the winning assumption. A supplier quote of $7.20 delivered for a notebook sold at $16 leaves $8.80 before fees, a 55% margin. The card fee at 2.9% plus 30 cents takes $0.76, so $8.04 is left before packaging and postage. Thirty notebooks a month would contribute $241 towards the tools in the budget table, and you can see in one line whether the idea is paying for itself or only for the subscriptions.
Weeks 1 and 2: A name from forty ideas and four checks
An assistant is good at volume here and useless as a judge. Ask for forty names against rules, then do the checking yourself.
Suggest 40 names for a stationery and craft-supplies shop. Rules: under 12 characters, easy to spell after hearing it once, no hyphens or numbers, works as a web address and an Instagram handle, nothing that sounds like a well-known brand. Group them by style. Do not tell me whether any name is available.The last sentence stops the assistant claiming a name is free, which it cannot know. Take the five you like, then run four checks and record them, as in this illustrative table:
| Name | Spelled from hearing | .com | Handle | Trade mark register | Verdict |
|---|---|---|---|---|---|
| A | 3 of 3 listeners | Free at $11.08 | Free | Similar mark in another category; ask an adviser | Shortlist |
| B | 1 of 3 | Taken | Free | Not checked | Drop |
| C | 3 of 3 | Listed at a four-figure premium price | Free | Clear | Drop: not worth the price |
| D | 3 of 3 | Free | Taken, one letter different | Clear | Maybe: handle confusion |
The spelling test costs three friends and a minute each. The domain check is a search at a registrar; one registrar lists a .com at $11.08 a year, at the same price on renewal, with free privacy. Compare renewal prices rather than first-year offers, because the second year is the one that surprises people. The handle check is the same search on every platform you plan to use. The register search belongs to wherever you trade, and a similar mark in a related category is a reason to ask a trade mark adviser before printing anything.
The assistant can do one more job: list what each shortlisted name means or sounds like in the languages your customers speak. Treat that as a prompt to ask a speaker, not a clearance. AI for startup branding covers voice and logo briefs once you have the name.
Choosing the store by what it costs per sale, not per month
Four options cover most one-person shops. The figures come from each vendor's pricing page in October 2026.
| Option | Subscription | Online card rate | In-person rate | Worth knowing |
|---|---|---|---|---|
| Shopify Basic | $39 monthly billing, $29 yearly | 2.9% + 30 cents | 2.6% + 10 cents | Its assistant, Sidekick, comes with the plans; POS Pro is an optional $89 a month per location |
| Square Free plan | $0 | 3.3% + 30 cents | 2.6% + 15 cents | The Plus plan, $49 a month, lowers the rates to 2.9% + 30 cents online and 2.5% + 15 cents in person |
| Wix Core | $29 a month on yearly billing, paid up front | Depends on the payment provider | Not listed | Core, Business and Business Elite accept payments; Light lists no payments. A Wix site cannot be moved to another host |
| Squarespace commerce plans (Basic, Core, Advanced) | Prices vary by currency; check the USD figure on its pricing page | Not compared | Not compared | A 2% store fee on Basic, none on Core and Advanced; a 14-day trial and no free plan |
The percentage matters less than it looks at small volume. At $32 an order, Shopify Basic's online card fee is $1.23 and Square Free's online rate is $1.36, a difference of $12.80 on 100 orders. Add the subscription and the picture changes with volume. At 20 orders a month, Shopify's $29 subscription is 54% of a $53.56 platform bill. At 200 orders the card fees of $245.60 are 89% of a $274.60 bill. So a new shop should compare subscriptions first, and revisit the card rates once orders pass a few dozen a month.
The cost of a wrong choice is the rebuild, not the monthly fee. Moving a store means re-uploading products, redirecting links and, for a builder that cannot export the site, starting again, so pick the platform you will actually open every week. If you sell from a stall as well, price the in-person rate separately: a weekend stall does not need Shopify's POS Pro, and Square's free plan handles it. Shopify also offers a free AI agent through its Inbox app on Basic and above, for stores using its new customer accounts, which answers customers' chats on its own. Test it on your own returns and delivery questions before you switch it on, because it can use web search as a secondary source and may answer from the wrong place.
Product pages and policies: AI drafts, the supplier sheet decides
Product descriptions are where an assistant saves the most time and where it invents the most. Give it only your supplier's facts and make it list what is missing:
Write a product page for an online shop using only the facts below. Plain English, 80 to 100 words, no superlatives, no claims that are not in the facts. If something a buyer would want to know is missing, list it after the description under MISSING.
FACTS: A5 dotted notebook, 192 pages, 120 gsm paper, lay-flat binding, 12 cover colours, supplier code NB-A5-D.An illustrative first draft, from a prompt without the facts-only rule:
Our A5 dotted notebook is perfect for bullet journaling, sketching and everything in between. With 192 acid-free, FSC-certified pages of premium 120 gsm paper, it is fountain-pen friendly and designed to last a lifetime. Choose from 12 stunning colours.
Four claims are not in your facts: acid-free, certified, fountain-pen friendly and lasting a lifetime. "Premium" and "stunning" are decoration. A version that survives the supplier sheet reads: "A5 dotted notebook with 192 pages of 120 gsm paper and lay-flat binding, in 12 cover colours. The dot grid suits bullet journals, lists and sketches." The MISSING list that the facts-only prompt produces is the more useful output, because it tells you what to ask the supplier: whether the paper is acid-free, whether it is recycled, and what colour the pages are. If you want to say anything about pens, test three fountain pens, two gels and a marker yourself, and write what you saw.
Use real photographs of real stock. If you use AI-generated images at all, some platforms require them to be labelled; Google's shopping listings, for one, ask for metadata marking AI-generated images. For descriptions across a whole range, writing Shopify product descriptions with AI covers the volume workflow.
The shipping and returns policy is the other page worth drafting with an assistant, and the one where each number must be a promise you will keep. Let it propose the structure: dispatch time, delivery options, the return window, who pays return postage, and what counts as faulty. Then replace every figure with your own. Consumer rights differ between places and an overpromise is expensive, so have the final wording checked before you publish it.
First customers in six weeks without an ad budget
With no ad spend, the plan is a sequence of small, countable actions. Each row below costs nothing in tools on the lean stack:
| Week | Action | Tool |
|---|---|---|
| 1 | A stall day to test products; collect email addresses on a paper sheet with a clear sentence about what you will send | Stall fee (ask the organiser), a phone camera |
| 2 | Photograph real stock and queue three posts a week | Canva Free, Buffer Free (three channels, ten queued posts) |
| 3 | Email 15 nearby businesses that have a gift counter or a waiting area, offering a sample set | Your email address |
| 4 | Send a launch email to the sign-ups | Mailchimp Free (250 contacts, 500 sends a month) |
| 5 | Claim the business listings that apply, such as Google Business Profile if customers can visit you | Free |
| 6 | Count visits, orders, sign-ups and stall sales; decide what to repeat | A spreadsheet |
The assistant is most useful for the writing. Give it the stock list with one true fact per item, and ask for a plan of posts:
Here is my stock list with one fact per item: [paste]. Write a four-week Instagram plan of 12 posts, three a week. For each post give the product, one real detail from the list, a caption under 40 words, and at most 5 hashtags. No claims that are not in the list. Tell me which photo to take.The five-hashtag limit is deliberate. Instagram has capped hashtags at five a post since December 2025, and assistants trained on older advice default to 15 or 30. One illustrative post from a reply: product, the A5 dotted notebook; photo, three notebooks fanned across a desk beside a pen; caption, "192 pages, 120 gsm paper, lay-flat. Dots, not lines. Which colour would you pick?" plus five hashtags. Check the detail against the stock list and the photo against the real stock before it goes in the queue.
For the outreach email in week 3, ask for a short message with placeholders and nothing invented:
Subject: Notebooks for your gift counter. Hi [first name], I run a small stationery shop and make up sets of three A5 dotted notebooks that suit a gift counter. I'd like to leave you a sample set at a time that suits you. If you like them, my terms for resale are [your terms]. Would that be all right? [your name]
The [your terms] placeholder is yours to fill, because the assistant has no way to know a margin that works for you. When customers are happy, ask them for a review in a plain sentence, with no reward and no picking who to ask: Google's review rules ban incentives and review gating.
Admin set-up: questions for an accountant, prepared with AI
The legal structure, registrations, sales-tax rules and insurance depend on where you trade and what you sell, which is why an assistant cannot do this part for you. It can make the meeting with an accountant shorter and better. Describe the shop in three sentences and ask for the questions you should bring, marking anything that depends on where you trade:
- Which legal structure suits a one-person shop selling online and at stalls, and what does each change about my personal liability and my tax?
- At what level of sales must I register for sales tax or its equivalent, and what do I charge before then?
- How should I record stock bought before launch, and can I claim it?
- How should I record stall sales paid in cash?
- Do I need a separate business bank account and card from day one?
- What insurance does a shop selling physical products usually need, and what does a stall organiser require?
- How long must I keep records, and in what form?
- Which accounting package suits my volume, given the limits on entry plans?
Treat the answers as the agenda for a paid conversation, not as advice. The last question is where the budget table matters. Xero's Early plan lists $27 a month and limits of 20 invoices and five bills, with auto-reconcile as a $5 add-on, so a shop buying from many suppliers can outgrow it quickly. QuickBooks Simple Start lists at $38 for one user. A spreadsheet with a bank export and a receipts folder is a respectable start if your accountant agrees, and AI bookkeeping for startups covers moving to software without losing the records.
Two smaller jobs belong in the same week. Turn on two-step verification for your email, store and bank, because the email account can reset all the others. And let the assistant draft a short privacy notice that says what customer data you collect and why, then have it checked, since the notice has to match what the shop actually does.
What to leave out for the first 90 days
The gap between the lean and comfortable stacks is $115.99 a month, or about $348 over a quarter. Seven lines make up most of it, each with a trigger for when it starts to pay, and until the trigger arrives the free tier is enough. The store runs the other way, because yearly billing saves $10 a month, which is why the seven lines below add up to $125.99 and not $115.99.
| Wait on | Monthly cost | Why it can wait | Buy when |
|---|---|---|---|
| A paid AI plan | $20 | Free tiers cover the idea test, names and first product copy | A usage limit stops real work on a named task |
| Business email on Workspace | $14 | A free address works at first, and Starter at $7 already includes Gemini in Gmail | Customers expect an address on your domain, or you use Gemini in Docs and Sheets every week |
| Canva Pro | $12 | Free includes up to 20 Standard or Premium AI uses a month | You run out of free uses or need several brand kits |
| Accounting software | $32 | A spreadsheet and a receipts folder cope while orders are few | Orders and supplier bills outgrow the sheet, or your accountant asks for it |
| Mailchimp Essentials | $13 | Free covers 250 contacts and 500 sends a month | You pass 250 contacts |
| Buffer Essentials | $15 | Free covers three channels and ten queued posts | You need a longer queue or more channels |
| Zapier Professional | $19.99 | Do the tasks by hand first; the free plan has 100 tasks and two-step Zaps | The same manual task repeats several times a week |
Before paying for a store at all, it is worth seeing what a free AI store builder gives a small shop. Billing frequency is a second decision with its own sum. Shopify Basic on yearly billing costs $348 up front against $468 over twelve months of monthly billing, which saves $120 if you keep the store all year. It also costs more than monthly billing for any shop that closes in the first nine months, because $348 divided by $39 is 8.9. Start monthly, switch to yearly when you are sure, and note that some annual purchases are non-refundable (Zapier's are).
The 90-day check, with an assistant auditing the subscriptions
At day 90, compare what happened with what the budget needed. The standard stack of $95.92 needs 8 orders a month at the sample margin. If you are taking 30, the stack is a rounding error and the real questions are stock and time. If you are taking 5, the tools are not the problem and cutting them will not fix it; go back to the idea test and ask which assumption failed.
Then audit the subscriptions. Export a month or two of bank lines, keep only the merchant, amount and date, and delete account numbers before pasting:
Below are my business card lines for the last two months: merchant, amount, date. Identify (1) any two tools that do the same job, (2) anything billed yearly and when it renews, (3) anything that looks like a free trial that has converted to paid. Do not guess what a charge is for if the merchant name is unclear; list it as UNKNOWN.A reply might flag two AI subscriptions at $20 each, a design tool renewing at $144 in three weeks and an automation plan billed at $19.99. The assistant cannot see usage, so you answer the one question that matters for each line: did it save an hour or earn money in the last 30 days? If not, cancel before the renewal date, and put each renewal date in your calendar when you buy.
The same discipline applies to the AI itself. A shop that started with the free tiers, wrote its pages from supplier facts, checked every claim and counted its first orders against a break-even figure has learnt more about its business than any generated plan would tell it. When the shop is trading, the next questions are pricing, stock and customer support, and each of them comes with its own tools and its own costs, so add them one at a time and run each through the same orders-to-cover sum.
Questions new shop owners ask about AI and start-up costs
Do I need a paid AI subscription to start?
Not at first. The free tiers of the main chat assistants cover the idea test, naming and first product copy. Pay for one plan when a usage limit stops real work, and pick one rather than three. Claude Pro lists at $17 a month on annual billing or $20 monthly, and the other big assistants price their main paid tiers at around $20.
Will AI shopping assistants such as ChatGPT show my products?
They can. ChatGPT suggests products where its shopping feature is available, with checkout on the seller's own store, and its Instant Checkout option is being phased out. Shopify's Agentic Storefronts, launched in March 2026, switch that discovery on by default for Shopify stores. There is no special trick: clear titles, accurate details and real photos help here as in ordinary search, and Google says its AI search features need nothing beyond normal search optimisation.
What if a tool the shop depends on changes its price or closes?
Plan for it from the start. Keep product facts, prompts, customer lists and order records in your own files, export them every month, and prefer tools with a clear export. Suppliers do change course: the calendar tool Clockwise shut down in March 2026 and deleted users' data rather than transferring it. Put renewal dates in your calendar and re-read the pricing page before each annual renewal.
How much should I budget for tools in the first three months?
For a one-person shop, about $40 a month on a lean stack built from free tiers, or about $96 on a standard one, before card fees, stock and rent. Introductory store offers lower the first quarter, but plan on the list price. At 90 days, compare the number of orders you took with the break-even orders in the budget table.
Further reads
- Starting a Business Using AI: A First-Month Plan for a Shop or Cafe — A first-month calendar for a shop or cafe.
- Can a Small Shop Use AI Without Having an Online Store? — If you will sell only in person at first.
- Taking a Local Shop Online With AI: Product Listings in a Weekend — Put a physical shop's range online in a weekend.
- How to Use Shopify Sidekick to Run Your Store Faster — What Shopify's built-in assistant can run for you.
- How to Buy AI Software for a Shop or Restaurant: Pre-Purchase Checks — Checks to run before paying for any AI software.
- Can AI Help a Small Shop Set Prices and Promotions? — Set prices and promotions once you have sales data.
- Your First 30 Days of AI Customer Support for an Online Shop — Handle customer questions in your first month of trading.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: vendor pricing pages read in October 2026 (Shopify, Square, Wix, Squarespace, Claude, Canva, Mailchimp, Buffer, Zapier, Xero, QuickBooks, Google Workspace and Porkbun). Prices change, so confirm each one before paying.