How Can I Use AI In My Business: A Practical Guide

Someone in your business is almost certainly using AI today. Probably ChatGPT, and probably for something sensible like tidying an email or summarising a long document. That’s a decent start. A plan comes next.

This guide covers how to use AI in your business in seven steps, from finding out what’s already happening to making the useful bits stick. It’s written for owners and directors of small businesses, so there’s no jargon and no hype.

The UK Government’s Business Data Survey found that 41% of UK businesses handling digitised data already use AI, and only 21% of those say their AI tools connect to the systems their businesses actually run on. Plenty of experimenting, not much joined up. The question most owners end up asking is: how can I use AI in my business without it turning into a mess?

1. Find out how AI is already being used

Before you buy any AI-powered tools, ask your team what they’re doing now. Four questions will get you most of the way:

• Which AI tools are you using?
• What are you using them for?
• Which tasks are saving you the most time?
• Are you putting company or customer information into them?

Expect some surprises. Some of the best ideas will come from people who have already found a shortcut with an AI assistant. Some of the worst will involve customer data or customer records pasted into a free account, which is how sensitive data ends up somewhere you can’t control. Our article on Shadow AI covers that risk in more detail.

2. Look for wasted time, not places to use AI tools

Asking “where can we use AI?” gets you a long list of vague ideas. Asking “where are we wasting time?” gets you a short list you can test.

Start with time-consuming tasks that nobody enjoys: data entry, follow-up emails, meeting notes, reporting, market research, and first drafts of blog posts and social media posts. Most of these are repetitive or routine tasks, and automating them is where AI tends to pay back the fastest. The real return is that it frees people for higher-value work.

Where AI can help depends on the business. In manufacturing, the wins are usually in the office around production, such as quality reports, supplier emails, inventory management and quoting from spec sheets. In professional services, it’s the first drafts of engagement letters and file notes. In financial services, it’s client letters and summaries of long documents, as long as client data stays inside tools you’ve approved. At a charity, grant applications and trustee reports consume the most staff time.

Some real-life examples of how small businesses get results: drafting customer communications, writing job descriptions, summarising long documents, helping you brainstorm ideas for marketing campaigns, and data analysis, such as asking it to analyse data in a spreadsheet before a decision-making meeting. Generative AI is good at first versions of all of these, and at content creation generally. It is not good at knowing your business, which is why a person still finishes the job.

Customer service chatbots are a popular first idea. They also put AI in front of your customers, so try AI-powered chatbots on internal questions first, and make sure customers can always reach a human. A bad chatbot costs you customer satisfaction faster than a good one wins it.

Pick a handful of tasks and test whether AI genuinely improves them. Plenty of ideas sound good and save nobody any time.

3. Set clear rules for staff

AI becomes a business risk when people aren’t sure what they’re allowed to do. Write down which tools are approved, what information can go into them, what must stay confidential, when AI-generated outputs need checking, and who is responsible for managing all of it.

Most businesses haven’t done this. The same survey found that only 17% of AI-using businesses have any kind of AI policy, and just 5% have a formal written one. A single page that your staff can read in two minutes puts you ahead of the majority. Protecting sensitive information should be the first line of defence. If your people handle personal data, the ICO’s guidance on AI and data protection is worth a read before you write it.

Good rules make experimenting safer. Banning it only pushes it out of sight.

4. Look at what you already have, then compare before you buy

If you use Microsoft 365, find out which AI features are already included or available as an add-on before you pay for anything new. Microsoft Copilot works inside Outlook, Teams, Word, Excel and PowerPoint, so it’s an obvious candidate. Obvious doesn’t mean best.

Plenty of other AI platforms and AI solutions are cloud-based services you can trial for a month, including Claude and ChatGPT. The only fair way to choose is to compare them on your own tasks. You don’t need technical skills for this, just a few real jobs and the same questions for each tool:

• How well does it do the tasks you actually have?
• What company data can it see, and where does that data go?
• How does it connect to the systems you already use?
• What does it cost per person per month?

Run the same few tasks through each one and see whose output you’d send to a client with the least editing. Licences are easy to buy and awkward to unwind, so it’s worth matching the tool to your business needs before you commit.

5. Start small

You don’t need an organisation-wide programme to implement AI. Choose one team and a handful of tasks, run them for a few weeks and ask four things: did it save time, was the output useful, did people actually use it, and were there any security or data concerns?

If the answers are good, build from there. If they aren’t, you’ve spent a few weeks and very little money finding out, which is a cheap lesson.

6. Keep a person responsible

AI models can produce useful drafts and analyses. They also make mistakes, and they make them confidently. A person needs to check anything important before it goes out and make a decision. Human oversight matters more as the stakes rise: a summary of your own meeting notes can get a quick glance, while a figure going into a board paper or a client contract needs to be checked against the source.

AI is only as good as the relevant data you give it, so data quality matters too. Messy spreadsheets give you messy answers, delivered in confident sentences.

7. Turn what works into the way you work

Once something works, don’t keep it a one-person party trick. Write down the task, the tool, the prompt you used and what a good result looks like. Showcase the people who do similar work, because supporting employees in using AI tools effectively matters more than buying more licences. Then ask whether the same approach would help another team.

Measure it against something real, such as hours saved or faster customer replies. That’s how adopting AI becomes part of how your business runs and a route to business growth, rather than something people occasionally try.

A simple place to start

Find out what’s happening now, pick where time is being lost, set some boundaries, test a few use cases, measure the results and build on what works. You don’t need to become an AI business overnight.

Want a second opinion?

We’ve supported 80+ businesses across Birmingham and the Black Country for over 20 years, and we use AI inside our own business. If you’d like help working out where to start, book a free 20-minute AI readiness review with James. We’ll look at how AI is used in your business today and agree on one or two sensible places to start. No preparation needed.