# How Should a Small Business Start Using AI?

Canonical URL: https://business.theaieducator.io/posts/how-should-a-small-business-start-using-ai
Publication: Dan Fitzpatrick Insights
Author: Dan Fitzpatrick
Topic: AI Strategy for Business
Published: 2026-09-22T22:32:18.000Z
Modified: 2026-09-22T22:52:24.963Z

Most small businesses have tried AI, and few have changed how the work gets done. Dan Fitzpatrick's answer: start with one job, not one tool. How to pick that job, the three rules to write first, and a 30-day plan for a small team.

## In brief

A small business should start using AI with one recurring job, not one tool: a task done every week that takes real hours and already gets checked before it leaves the building. Give it a named owner, write three rules (what AI is for, what staff may do with it, who decides when it goes wrong) and run it for 30 days before expanding.

## Key takeaways

- Most small businesses have tried AI but few have changed how work gets done: in a Goldman Sachs survey published in March 2026, 76% of owners said they use AI and only 14% said it is fully embedded in their core operations.
- Access is rarely the barrier; the UK Office for National Statistics lists difficulty identifying business use cases among the factors most commonly reported as preventing or delaying AI adoption.
- The One Job Rule says to start with one recurring job, not one tool, and make AI part of that job for 30 days before buying, rolling out or announcing anything else.
- A good first job passes three tests: it is weekly, it is weighty, and a person already checks the output before it leaves the building.
- Before anyone starts, three rules on one page are enough: what AI is for here, what staff may do with it, and who decides when it goes wrong.
- Final client advice and decisions about a person's job, health, money or legal position are poor first jobs for AI in a small business.
- Once three jobs are working, the next step is a whole-business view of AI readiness and strategy rather than more single experiments.

In a Goldman Sachs survey of small-business owners published in March 2026, 76% said they use AI. Only 14% said it is fully embedded in their core operations. Most owners have opened the door. Few have walked through it.

Here is the answer I give owners who ask where to begin. Start with one job, not one tool. Pick a single task that someone in your business does every week, that eats real hours, and that a person already checks before it leaves the building. Give it a named owner, write three plain rules about what may go into AI, and make AI part of that job for 30 days. Then let what you learn choose the second job. Strategy, a full policy and company-wide training all come easier once one job is working, and none of them can stand in for it.

That answer holds whether you run a 12-person agency in Manchester or a 90-person distributor in Ohio.

## Why Do So Many Small Businesses Start in the Wrong Place?

Most small businesses start with a product or a plan, when the thing that has to change is a piece of work someone does on a Tuesday afternoon.

Official statistics on both sides of the Atlantic point the same way. In the UK, the Office for National Statistics reported in [Artificial intelligence in UK businesses: 2023 to 2026](https://www.ons.gov.uk/businessindustryandtrade/business/businessservices/articles/artificialintelligenceinukbusinesses/2023to2026) (July 20, 2026) that self-reported AI use among businesses with 10 or more employees has risen "from around 12% to around 35% since late 2023," but calls that adoption "relatively shallow." Among the factors most commonly reported as preventing or delaying adoption, the ONS lists "difficulty identifying business use cases."

In the US, the Census Bureau's [May 26, 2026 analysis](https://www.census.gov/library/stories/2026/05/ai-use-businesses.html) of its Business Trends and Outlook Survey found that "overall AI usage hovered between 17% and 20%" from December 2025 to May 2026. Use rose among firms with at least 20 employees but "didn't change significantly among firms with fewer than 20 employees." At the small end: "Less than 20% of firms with four or fewer employees reported using AI," against 37% of firms with at least 250 employees.

Those figures do not contradict Goldman Sachs. They measure different people. The [Goldman Sachs survey](https://www.goldmansachs.com/pressroom/press-releases/2026/small-businesses-embrace-ai-but-need-training-and-support-to-fully-harness-it) (March 17, 2026) asked 1,256 owners taking part in its 10,000 Small Businesses program, people who have already signed up to grow their business. The Census asks a broad sample of US firms whether they use AI "in any business function." Read together, they suggest that owners who pay attention have tried AI, far fewer have changed how the work gets done, and the smallest firms are standing still.

My interpretation: for small and medium-sized businesses (SMBs, or SMEs in the UK and Europe), the barrier is rarely access. Anyone can open a chatbot tonight. The barrier is knowing which piece of work to point it at. The UK government's SME Digital Adoption Taskforce, in its [2026 update](https://www.gov.uk/government/publications/sme-digital-adoption-taskforce-2026-update/sme-digital-adoption-taskforce-2026-update) (June 26, 2026), names three barriers: "capability, cost and awareness." A first job built into someone's week tackles the first and the last without a large project. If you have already tried AI and nothing has stuck, I look at the causes in [why most small businesses are not getting value from AI yet](https://business.theaieducator.io/posts/why-small-businesses-are-not-getting-value-from-ai).

## What Is the One Job Rule?

The One Job Rule swaps the question "Which AI tool should we use?" for a better one: "Which job should AI join first?"

This is the definition I use with owners:

> **The One Job Rule** is the way I suggest a small business starts with AI: choose one job, not one tool. Pick a single recurring task that happens every week, takes real hours, and already gets checked by a person before it leaves the building. Give it one named owner, and make AI part of that job for 30 days before you buy, roll out or announce anything else.

Three tests pick the job. I call them weekly, weighty and checked.

- **Weekly.** The job comes around at least once a week, so the person doing it gets enough repetitions to learn where AI helps and where it stumbles. A task you do twice a year teaches you nothing.
- **Weighty.** It takes enough time that saving part of it matters. Twenty minutes a month is not worth changing a habit for.
- **Checked.** A person already reviews the output before it reaches a customer, a supplier or a regulator. AI will get things wrong, and a check that already exists catches the mistakes while your team learns.

The named owner matters as much as the job. "The office" cannot learn a new way of working. Your office manager can.

The rule also carries the principle I apply to every use of AI at work: outsource the doing, not the thinking. In the first job, AI drafts, sorts or summarizes. A person still judges.

## How Do You Choose the First Job?

To choose the first job, list every recurring task that crosses a desk in an ordinary week, then keep only the ones that are weekly, weighty and checked.

| Usually a good first job | Usually a poor first job |
|---|---|
| First drafts of quotes, proposals and bids | Final advice to a client |
| Replies to routine customer emails, reviewed before sending | A chatbot that answers customers with nobody reviewing it |
| Turning meeting notes into a list of actions | Hiring, pay or disciplinary decisions |
| Job ads and product descriptions | Anything that decides something about a person's health, money or legal position |
| Monthly management commentary drafted from figures a person has checked | Tasks that come around once a quarter |

Everything in the right-hand column either fails one of the three tests or hands AI a judgment that belongs to a person. Some fail "weekly." The dangerous ones fail "checked": the error would leave the building before anyone saw it.

Take a hypothetical 22-person heating and plumbing firm. The owner's first instinct is a chatbot on the website. It fails the checked test: nobody reviews what it tells a customer at eleven at night. Look at the week instead. The operations manager spends most of Friday turning engineers' site notes into written quotes. That job is weekly, it is weighty, and every quote is already checked before it goes out. It is the first job. For four weeks she drafts each quote with AI from the site notes, checks it as she always has, and writes down two things: the time each quote took and every error she caught. At the end of the month the owner holds something no sales demo can supply: evidence from their own business.

## What Rules Does Your Team Need Before Day One?

A small business needs three rules on a single page before anyone starts: what AI is for, what staff may do with it, and who decides when it goes wrong.

I build them from the Readiness Test: "The Readiness Test is three questions an organization must be able to answer consistently, from the top to the front line, before it can call itself AI ready: What is AI for here? What may I do with it? Who decides when it goes wrong?"

For a first job, the answers can be short:

1. **What AI is for here.** "We are using AI to draft quotes from site notes. That is the only approved use for now."
2. **What you may do with it.** Name the approved tool and insist on a business account rather than personal ones. Name the information that never goes in: customers' personal details, bank details, anything covered by a confidentiality agreement.
3. **Who decides when it goes wrong.** One named person to tell when something looks off, and a promise that nobody gets blamed for raising it.

That is not a full policy. It is enough to start safely, and it gives the full policy real experience to draw on. For when you need the longer version and what belongs in it, read [does a small business need an AI policy?](https://business.theaieducator.io/posts/does-a-small-business-need-an-ai-policy)

## What I See When I Train Business Teams

Across the business teams I train, the first question is almost always "Which tool should we buy?" The businesses that get furthest ask a different one: which job?

I have trained more than 150,000 people across more than 30 countries, and the pattern holds across sectors. Teams that learn AI in the abstract nod along and return to their inboxes. Teams that learn it on a job they did last Thursday change what they do next Thursday.

The clearest business example I can share publicly is theidol.com. In April 2023 I spoke about AI in marketing at a business event in Penrith. A year later, the company's marketing team was using AI for content writing. Then the managing director asked me back to show teams across the business what was possible. After that session, in September 2024, the IT Director wrote: "It was a great session and I've heard a lot of good feedback from the attendees."

Notice the order. One team used AI on one kind of work first; the wider business followed. Nothing began with a company-wide launch.

Most businesses have not yet trained their people in any depth. The ONS reported: "Only 11% of businesses (with 10 or more employees) reported that more than half of their workforce has received AI-related training." In the Goldman Sachs survey, 73% of owners said they "would benefit from additional access to training and resources." Training sticks when it is anchored to a real job, which is why the first job and the first training session should be the same thing. I set out what works in [AI training for small business teams](https://business.theaieducator.io/posts/ai-training-for-small-business-teams).

## What Should You Do in the First 30 Days?

The first 30 days should cover one job and one owner, in four steps: set up, use, share and decide.

- **Week 1: choose and set up.** List the recurring tasks, apply the three tests, pick one job and name its owner. Write the three rules. Record how long the job takes today; without a "before" figure, you will argue about the "after."
- **Week 2: do the job with AI.** The owner uses AI on every instance of the job and keeps a short log: time taken, what AI got wrong, what they changed.
- **Week 3: show one colleague.** The owner walks one other person through how they now do it. If they cannot explain it in fifteen minutes, it is not a habit yet.
- **Week 4: decide.** Keep, adjust or stop. Then choose the second job, with a different owner.

Before you start, put your plan through the Monday Test: "could a member of staff read it and know what to do differently on Monday morning?" If the owner of your first job could not, the plan needs another pass.

If you do one thing this week, block thirty minutes on Friday afternoon, list every task your team repeated since Monday, and circle the ones that are weekly, weighty and checked. Your first job is almost certainly on that list.

## When Is One Job Not Enough?

The One Job Rule is a starting rule, and three situations call for more.

**Your staff are already using AI on their own.** Then the rules come first, this week, because client information may already be going into tools nobody has approved. Choose the job once the page of rules exists.

**You work in a regulated field.** Law, accountancy, financial advice, healthcare: you can still start with one job, but whoever holds your compliance responsibilities, in the US or the UK, should sign off on the three rules before the first draft is written.

**Your first three jobs are working.** Now you need a strategy: which parts of the business matter most, who owns AI, and how you will know it is paying off. This is the moment to look at readiness across the whole business. My Seven Dimensions of AI Readiness cover purpose, permission, ownership, capability, protection, foundations and evidence. As I put it: "Your readiness is your lowest score across the seven, not your average." I explain each one in [how ready is your business for AI?](https://business.theaieducator.io/posts/how-ready-is-your-business-for-ai)

The one-job start is not a small ambition. It is how the ambitious version gets built without betting the business on a guess.

## Where to Go From Here

If you want to know where your business stands before you choose that first job, take the free [Workplace AI Readiness Check](https://theaieducator.io/workplace-ai-readiness?utm_source=business.theaieducator.io&utm_medium=referral&utm_campaign=how-should-a-small-business-start-using-ai): 21 questions, about eight minutes, scored across seven dimensions. If your leadership team needs to agree on what AI is for before anyone starts, that is what my [AI strategy and governance work](https://theaieducator.io/ai-strategy-and-governance?utm_source=business.theaieducator.io&utm_medium=referral&utm_campaign=how-should-a-small-business-start-using-ai) is built for, beginning with a half-day session built on the Readiness Test.

## Sources and further reading

- [Small Businesses Embrace AI But Need Training and Support to Fully Harness It](https://www.goldmansachs.com/pressroom/press-releases/2026/small-businesses-embrace-ai-but-need-training-and-support-to-fully-harness-it), Goldman Sachs 10,000 Small Businesses (press release; survey of 1,256 participants, January 27 to February 4, 2026), March 17, 2026.
- [Large Firms With at Least 20 Employees Biggest AI Users](https://www.census.gov/library/stories/2026/05/ai-use-businesses.html), US Census Bureau (Business Trends and Outlook Survey), May 26, 2026.
- [Artificial intelligence in UK businesses: 2023 to 2026](https://www.ons.gov.uk/businessindustryandtrade/business/businessservices/articles/artificialintelligenceinukbusinesses/2023to2026), Office for National Statistics, July 20, 2026.
- [SME Digital Adoption Taskforce: 2026 update](https://www.gov.uk/government/publications/sme-digital-adoption-taskforce-2026-update/sme-digital-adoption-taskforce-2026-update), Department for Business, Innovation, Science and Trade and Department for Business and Trade (GOV.UK), June 26, 2026.

*[Dan Fitzpatrick](https://theaieducator.io/about) is The AI Educator: a Forbes contributor, bestselling author and international keynote speaker who has trained more than 150,000 people across 30+ countries.*


## Frequently asked questions

### What is the first step for a small business to start using AI?

The first step is choosing one recurring job, not a tool. Pick a task someone does every week, that takes real hours and already gets checked before it leaves the building. Give it a named owner, write three simple rules, and make AI part of that job for 30 days before expanding.

### What should a small business use AI for first?

Use it first on routine drafting that a person already reviews: quotes, proposals, replies to routine customer emails, job ads, meeting actions or monthly commentary drafted from checked figures. Avoid anything that reaches customers unreviewed or decides something about a person's job, health, money or legal position.

### Does a small business need an AI policy before staff start using AI?

You need rules before anyone starts, but not necessarily a long policy. One page answering three questions is enough to begin: what AI is for here, what staff may do with it, and who decides when it goes wrong. Write the fuller policy once you have real experience to draw on.

### How many small businesses are using AI in 2026?

It depends on who is asked. The US Census Bureau reported in May 2026 that less than 20% of firms with four or fewer employees used AI. The UK Office for National Statistics reported in July 2026 that 28% of businesses with 0 to 9 employees used at least one AI technology.

### How long does it take a small business to see whether AI is working?

Start with one job and you can have evidence within a month. Record how long the job takes before you begin, then log the time taken and every error caught for four weeks. That gives you a before-and-after comparison from your own business rather than a vendor's promise.

### Which AI tool should a small business choose first?

Choose the job before the tool. Many small businesses start with a general assistant such as Microsoft Copilot or ChatGPT, used on a business account rather than personal ones. The right tool is the one that fits your first job and keeps client information under your control.

### Is it safe for staff to put customer data into AI tools?

Not by default. Decide which approved tool and account your team may use, and list the information that never goes in, such as customers' personal details, bank details and anything covered by a confidentiality agreement. In regulated fields, whoever holds compliance responsibility should sign off those rules first.

---
Source: [How Should a Small Business Start Using AI?](https://business.theaieducator.io/posts/how-should-a-small-business-start-using-ai)
Publisher: [The AI Educator](https://theaieducator.io)