Before you bring in AI, do this first.

Pick one goal and one owner

Why it matters. “Use AI” isn’t a plan. One clear goal tells you what to measure and what it’s worth.

What to do

• Pick one problem that costs you time or money, like slow invoices or missed enquiries.
• Say what you want to improve and how you’ll measure it. For example, cut invoice time from 10 minutes to 2.
• Name one person to own it. Not a committee.

You’re ready when you can name the job, the measure and the owner.

Map the job and cut the waste

Why it matters. Automating a messy job just gives you a faster mess. A simple map shows what’s worth keeping.

What to do

• Write down what comes in, like an email or a call, and what goes out, like an invoice or a booking.
• Ask the people who do the job how it really works today.
• Cut extra steps, repeated work and retyping before you automate anything. Keep the checks and approvals that matter.

"Don't automate, obliterate."

Michael Hammer, Harvard Business Review, 1990

You’re ready when a new starter could follow the job, and it’s simpler than before.

List the systems the job uses

Why it matters. AI works with the information it can reach. Most businesses use more apps and spreadsheets than they think.

What to do

• List every system and spreadsheet the job touches.
• Note where information starts and where it gets copied.
• Note which system holds the information you need, and whether it can share it securely.
• Start with the job from step 1, not the whole business.

You’re ready when you have a simple picture of your systems and how information moves between them.

Get your systems sharing

Why it matters. When the same customer is typed into three places, the copies can drift apart. Knowing which system to trust for each kind of information keeps everyone, AI included, on the same page.

What to do

• For each kind of information, pick the system you trust most. For example, your CRM for customers and your accounts software for invoices.
• Connect systems where you can, so information is typed in once.
• Where they can’t connect, agree how you’ll keep them in step.

You don’t need every system linked before you start. Just the ones your chosen job uses.

You’re ready when you know where reliable information comes from, and how the systems in your chosen job work together.

Check your data and measure where you are today

Why it matters. Measure the job today and you’ll be able to show exactly what improved.

What to do

• Check the information you rely on is accurate, complete and free of duplicates.
• Pick one measure, such as hours a week or days to send an invoice.
• Write down today’s number. That’s your starting point.

You’re ready when you can show today’s number, and the team agrees with it.

Agree simple rules for data and AI

Why it matters. Clear rules let everyone use AI with confidence. UK data protection law still applies to personal data.

Start this one early, alongside step 1. Don’t leave it until halfway through.

What to do

• Write a one-page policy: which AI tools to use, and what information can go into them.
• Give each AI tool access only to the information and systems it needs.
• Ask each supplier where your data is kept and how you’d get it back if you leave.
• If personal data is involved, check the ICO’s guidance (linked below), including whether you need a data protection impact assessment.
• Decide what AI can do on its own and what needs a person’s sign-off, especially money, personal data and important decisions.

You’re ready when everyone knows which tools to use, what they can access, and when a person signs off.

Bring your team in

Why it matters. The people who do the job know its quirks. Involve them early and they’ll help shape something that works.

What to do

• Involve them from the start, not on launch day.
• Ask which part of the job they’d hand over first.
• Ask what could trip the new way up, and what they’d need to rely on it with confidence.
• Give them time to learn the new way of working.

You’re ready when the team can explain their part in the new way of working.

Try it on one job first

Why it matters. A short trial on your own work shows what AI can do for you, at low cost.

What to do

• Pick one job, one team and a few weeks.
• Agree up front what success looks like.
• Test it with real examples, including unusual cases and the ones it should pass to a person.
• Keep the old way running alongside until you’re sure.

You’re ready when the trial has clear measures, has been tested on real examples, and can be paused safely if needed.

Measure, then decide

Why it matters. Compare against your starting point and let the results choose the next step.

What to do

• Compare time saved, accuracy and running costs with your starting number.
• Ask the people using it what they think.
• Roll it out, adjust it, or try it on a different job.

You’re ready when the decision is based on results, not hunches.