AI is in use today in nearly every industry, and in small firms it is mostly doing four things: answering routine questions, drafting text, sorting documents and planning schedules. Retail, trades, health, finance, logistics, hospitality and education each have one job where it pays first, usually costing between nothing and about $150 a month.
Overviews of this subject tend to list AI's grandest uses, such as vision systems and autonomous vehicles, which have little to do with a firm of five. What follows is smaller and checkable. For each industry, one job a small firm actually gives to AI, what it costs, a sample of the work, and the check that keeps a mistake from reaching a customer. Figures are list prices where stated and illustrative where not.
Retail: answering the same ten shopper questions
An online shop or a small chain's web store gets a steady flow of questions that sit in a policy page somewhere: where is my order, what is your returns window, does this come in blue, is it in stock. The small-firm job is an AI agent that answers those from the shop's own information and passes everything else to a person.
The cost can be close to nothing. Shopify's Inbox agent is opt-in and the Inbox app is listed as free, and Tidio's free plan includes a one-time allowance of 50 conversations with its Lyro AI agent before paid tiers start. Check each vendor's page for the plan requirements, since they change.
A sample: a customer writes "Can I still return the jacket I bought three weeks ago?" The agent replies, from the policy page, "Returns are accepted within 30 days of delivery if the item is unworn with tags. Yours was delivered 18 days ago, so you're inside the window. Reply with your order number and we'll send a label." The check is a weekly read of 20 conversations. The owner looks for one error that matters more than all others: a promise the policy doesn't make. Rule to add: the agent never offers refunds, discounts or exceptions. Compare the options on cost per conversation at your own volume, not on the free allowance.
The sum: a shop that gets 60 routine questions a week at 3 minutes each spends 3 hours answering them. If the agent handles two-thirds, that frees 2 hours a week, about 8 to 9 hours a month, which at $20 an hour is $170 of time for a tool that costs nothing to start.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowTrades: the missed call and the quote follow-up
A plumbing, electrical or heating firm with five vans loses work when nobody answers the phone during a job. The small-firm job is an AI receptionist that answers, takes the details and books or flags the call. A related and easier job is the quote follow-up: a message sent three days after the quote to ask if the customer has questions.
On cost, Jobber's help centre lists its Receptionist as a $29-a-month add-on on selected plans, covering 30 conversations, with $0.79 for each conversation after that, and the Plus plan includes it with unlimited use. Older pages have quoted other allowances, so confirm the figure on your own account before you rely on it. Other field-service platforms have their own receptionists with prices that differ, so compare on cost per conversation.
A sample call summary the office sees: "Caller: [first name]. Job: boiler not heating, pilot light out, no gas smell reported. Property: two-storey house. Availability: tomorrow morning. Flagged: URGENT-SAFETY NOT TRIGGERED." The check is a rule that any mention of gas smell, smoke, flooding or sparks goes straight to a person's phone and tells the caller to follow your emergency advice. Jobber's version lets you choose the keywords that escalate a call or text to a named team member, so set those first. Run a test with ten made-up calls before it takes a real one. For a fuller costing by van count, read what a small plumbing firm pays each month.
The sum: if the firm misses six calls a week during jobs, and one in three of them was a job worth $250, a receptionist that recovers half the missed calls earns back roughly $250 a week, against $29 plus conversation charges. That works if the calls are real jobs, and fails if most are sales calls, which is why the test calls matter.
Health: clinic notes that don't eat the evening
A small clinic's clinicians spend a surprising share of their day writing notes. The small-firm job here is an AI scribe: it listens to the consultation, with the patient's permission, and drafts the note for the clinician to edit and sign. It is a good fit because the clinician reads every line, the output is text, and the time saved is large.
The cost range is wide, and the examples vary by platform. Jane lists its AI Scribe at $15 per opted-in practitioner a month for unlimited notes, or five free notes a month to try it, while Heidi, a stand-alone scribe, lists its Clinician plan at $110 per user a month billed annually alongside a free plan. Check the pages for the plan you would use.
A sample: a physiotherapy consultation of 25 minutes becomes a draft note with the presenting problem, findings, treatment given and plan. The clinician finds that the AI wrote "pain improved to 3/10", when the patient said "about a 4", and fixes it. That is the check in miniature: the clinician reads every figure and every clinical statement before signing, because the clinician, not the tool, is responsible for the record.
Patient data brings extra steps: patient consent for recording, a business-grade product with a clear data-use policy, and your regulator's rules on records and AI. AI scribe costs compared for small clinics compares the options. The sum: if a clinician saves six minutes a note across 18 notes a day, that is about 1.8 hours a day.
Finance: month-end commentary and document capture
For an accountancy or bookkeeping practice, two jobs lead. The first is capturing bills and receipts into the ledger. Xero's Smart Document Capture, launched in July 2026, is free whatever the small-business plan and asks you to approve the details before it creates a record. Dext prices practices per client, from $17.70 a client a month on an annual subscription with a ten-client minimum, and lists $302.50 a year (250 documents a month, five users) for a single business. The second is drafting commentary and client chasers from data a person has already reconciled.
A sample of the second job: from a list of five unmatched bank items, the assistant writes a client message that names each date and amount and asks for a receipt or explanation. The preparer checks every figure against the ledger before it goes. The risk to manage is confidentiality: use a business plan, mask client names before pasting, and add a line to your engagement letter. Month-end tasks to hand over first goes through that in detail.
The sum: a practice with 40 clients that saves 20 minutes per client each month recovers about 13 hours. At $45 an hour that is $600 against tool costs that start under $100.
Logistics: a morning route plan and delay messages
A courier or small haulage firm with ten vans needs routes built each morning and customers told when a van is running late. The small-firm job is a route planner that builds the day from the order list, with AI drafting the delay messages for a dispatcher to approve.
Cost: Routific lists a free tier for up to 100 orders a month and a flat $150 a month for up to 1,000, with per-order bands above that. Other planners price differently: Spoke Dispatch, for example, shows a $125-a-month Starter plan with 1,000 stops included on its pricing page.
A sample message, drafted from the stop's data: "Hi [first name], your delivery is running about 40 minutes behind. Your new window is 12:30 to 14:30. The driver will call 15 minutes before arriving." The checks are on the estimate: compare planned and actual arrival times for two weeks, and fix the largest cause of error first, which is often the time allowed at each stop. Modelling a full day is in dispatch and delay alerts for 10 vans. The sum: if 14 of 180 daily stops run late and each used to cost a 4-minute phone call, proactive messages that halve those calls save about half an hour a day.
Hospitality: stock, rotas and review replies
A café, small restaurant or guest house has three recurring jobs that suit AI: replying to reviews, turning last month's sales into a prep and order plan, and drafting the staff rota from forecast covers. Review replies are the easiest. An owner pastes a review and asks for a polite, specific reply in the house voice; the owner reads it and posts it.
Cost is low. Many restaurant point-of-sale systems include a basic assistant; Toast IQ, for example, is listed as included with Toast POS, and a general assistant on a personal plan is about $20 a month. A sample reply to a three-star review about a slow lunch: "Thank you for telling us, [first name]. Lunch on a busy Saturday should not take 50 minutes, and we're sorry it did. We've added a second person to the pass at weekends. We'd like to make it up to you, so please ask for the manager on your next visit." The check: the owner confirms that the claim (the second person on the pass) is true before the reply goes out, because a reply that promises a fix that isn't there makes things worse.
The sum: 20 reviews a month at 8 minutes each is under 3 hours; at 3 minutes each with AI, it is an hour. The larger win in hospitality is forecasting prep, covered in small hotel and café jobs for AI.
Education: parent reports and lesson prep
A small tutoring centre or independent teacher spends evenings writing progress reports and building worksheets. The small-firm job is drafting both from the teacher's own notes. For reports, the tutor supplies three facts per pupil (topic covered, one strength, one next step) and the assistant writes a short paragraph in a consistent voice. For worksheets, it drafts questions at a stated level with an answer key.
The cost is that of a general assistant, about $20 a month for a personal plan or $20 to $25 a seat on a business one, plus any design tool allowance. A sample: from "fractions, confident with halves and quarters, needs practice with unlike denominators", the draft report reads, "[First name] has made good progress on fractions this term and is confident working with halves and quarters. The next step is adding fractions with different denominators, which we'll practise weekly." The check is that every worksheet is solved by the tutor before use, as AI-made answer keys have errors. Pupil information stays out of the tool unless the plan and your own policies allow it. See how tutoring centres write parent progress reports with AI.
The sum: 30 pupils' monthly reports at 10 minutes each is 5 hours. At 4 minutes each with review, it is 2 hours, a saving of 3 hours a month.
What the seven have in common
| Industry | The small-firm job | Typical monthly cost | Checked by |
|---|---|---|---|
| Retail | Answering routine shopper questions | $0 to start | Owner reads 20 chats a week |
| Trades | Answering calls and following up quotes | About $29 plus usage | Office reads call summaries, tests emergency rules |
| Health | Drafting consultation notes | $0 to $110 per clinician | Clinician edits and signs |
| Finance | Document capture and client commentary | From about $40 for two assistant seats | Preparer checks figures to the ledger |
| Logistics | Routes and delay messages | $0 to $150 and up with volume | Dispatcher approves, compares ETAs to actuals |
| Hospitality | Review replies and prep planning | About $20 | Owner confirms every claim |
| Education | Parent reports and worksheets | About $20 to $25 | Tutor solves worksheets, reads reports |
In every row, the AI does a draft, and a named person with knowledge of the subject checks it before anyone outside the firm sees it. That is the real pattern behind AI use in small firms, and it holds across all seven. The tools change often, and prices move, so use the table to see the shape, then check each vendor's page before you buy.
The large-company uses you can ignore for now
Much of what gets written about industry AI describes systems that a firm of five will never buy: computer vision that inspects every item on a fast production line, models that forecast demand across hundreds of sites, robots in distribution centres, and clinical tools that read scans. They are real, and they are expensive to set up and need specialist data. They also depend on volume. A line that makes a million parts a month justifies a vision system; a shop that makes four hundred does not.
A good test for any idea from a big-company case study is to ask what it needs that you don't have. Is it years of clean data, a technical team, a contract measured in tens of thousands, or a volume of transactions you only reach on your best day? If so, park it and take the smaller version: the same idea applied to a spreadsheet export and a general assistant. A shop that cannot afford demand forecasting software can still paste three months of sales into an assistant and ask which products sold out twice. The result is cruder, and it is also a great deal cheaper and quick to try.
A short checklist before you pay for anything
- Is the price per user, per site, per conversation or per document? Work out the monthly figure at your real volume, not the plan's headline.
- Does the plan say your content isn't used for training? Business plans usually say so by default; personal plans often depend on a setting.
- Can you try it free, or for a month, on real work before committing to a year?
- Who in your firm is the named checker, and is it written into their week?
- What happens if you stop? Can you export your data, and does the work carry on without the tool?
Five yes-or-no answers take ten minutes and filter out most of the products that look good in a demonstration and cost more than they save.
The chatbot that promised a refund
A realistic failure from the retail row: a shop turns on an assistant to answer customers and gives it the policy page. A customer writes that a mug arrived chipped. The assistant, trying to be helpful, replies that a full refund will be issued within two days. The policy actually says damaged goods need a photo first, and the owner reads the transcript the next morning to find that 14 such promises went out that week. Eleven were fine anyway. Three were for items that the customer had broken after delivery.
The lesson applies to every industry in the table: rules on what the AI may promise matter more than how well it writes. Put them in writing ("never promise a refund, a price or a time"), test them with awkward questions before launch, and review real transcripts weekly for the first month. The same habit prevents a trades receptionist from quoting a price or a scribe from adding a diagnosis the clinician never made.
Choosing which row to start with
Take your own firm through four questions and the first job usually becomes clear.
- Where do staff repeat themselves? The same question answered, or the same message typed, ten times a week is the best first target.
- Can a mistake be caught before it leaves? Prefer jobs where a person reads the output first.
- Does it need your private data? If yes, choose a business plan and a masking rule before you start. If not, start there, because it's simpler.
- Can you measure it? Count minutes or messages for a week now, so that you can compare in a month.
Start with one job, run it for a month, and only then add a second. A firm that does one thing well with AI learns more than a firm that tries seven. If you'd like help finding the first job in your own business, that is what my AI implementation consultation covers.
Questions about AI use across industries
Which industry gets the most from AI for a small firm?
There is no single winner. The best returns come from any industry where staff spend hours on repeat writing, answering the same questions or sorting documents. Accountancy, trades offices, clinics with note-heavy days and online shops tend to see quick results because their work is text-based and the outputs can be checked.
Are some industries too risky for small firms to use AI?
Regulated work such as health and finance needs more care, because client or patient data and professional duties are involved. It isn't off limits. It means business-grade plans, a written rule for what may be pasted, a named person who checks outputs, and an eye on your regulator's guidance before you start.
How much should a small firm expect to spend?
Most of the examples here cost between nothing and about $150 a month per user or per site. A general assistant on a business plan is around $20 to $25 a seat, and several industry tools have free tiers or small add-ons. The larger cost is staff time to set rules and check output in the first month.
Should I buy an industry-specific tool or use a general assistant?
Begin with a general assistant if the job is writing or summarising, because it is cheap and flexible. Move to an industry tool when the job needs a connection to your own system, such as a booking or accounting platform, or when it must meet a record-keeping standard that the general tool does not.
Further reads
- How to Build a Restaurant Staff Rota With AI Demand Forecasts — Build a restaurant rota from demand forecasts.
- Is AI Worth Paying For in a Small Tutoring Business? — Whether AI pays for itself in a small tutoring business.
- Best AI for a Boutique: Tools Compared by Job and Monthly Cost — Boutique AI tools compared by job and cost.
- What an AI Implementation Looks Like in a Small Accounting Firm — What an AI roll-out looks like in a small accountancy.
- Can a Small Shop Use AI Without Having an Online Store? — How a shop with no website can still use AI.
- How Dental Practices Use AI for Recalls and Appointment Reminders — How dental practices use AI for recalls and reminders.
- Best AI Chatbots for Small Shopify Stores (2026) — Chatbot options for a small online shop, compared.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Shopify Inbox help and app pages; Tidio pricing; Jobber Receptionist help article; Jane and Heidi pricing pages; Xero Smart Document Capture page; Dext business and practice pricing; Routific and Spoke Dispatch pricing; Toast IQ product page; OpenAI and Anthropic plan pages; illustrative figures throughout.