Whether or not you've approved AI, it's probably already in your business.
Some leaders are all in. Some are cautious, for good reasons. Either way, AI is usually doing more of your marketing than the last progress update said, and the work around it is easy to miss from the top. This page sets out what is worth knowing, from your side of the table.
Whoever sets the direction or signs off the budget. In some businesses that is one owner. In others it is a board, a group of partners, or a director who has recently taken marketing on.
If you lead a team and report to someone above you, the middle-seat page is written for you. If there's no marketing team at all, try the owners' page.
Four things worth knowing.
- The gains are real. More output, faster, for less. Used well, AI is a good investment.
- The work has moved. Less time producing, more time checking, correcting and approving. Measure it once in hours, and capacity planning gets much easier.
- Knowledge sits with very few people. Often one person understands how your AI set-up fits together. Writing it down protects the business, and them.
- Ownership needs a name. Decide who signs off AI work before customers see it. Clear ownership is what lets a team move quickly without worry.
of UK workers who use AI at work have relied on its output without checking how accurate it was.KPMG Trust in AI study, UK respondents, 2025
A safe way to start.
Caution is reasonable. It is also worth knowing that most staff who use AI at work bring their own tools, so the question is often how to bring existing use into the open.
- Find out what is already in use. Ask openly, with no blame attached. Include free tools on personal accounts.
- Set two rules on day one. No personal data (customers', clients' or staff) and nothing confidential in AI tools without approval, and nothing goes to a customer without a person reading it first.
- Run one small trial. Choose one low-risk job, such as first drafts of internal reports, on a business account, for one month.
- Name who signs off. One person, by role, for anything AI-produced that leaves the building.
Getting more from it.
The businesses that get the most from AI are the ones that can see the whole picture. That means the checking time as well as the output, the cost as it grows, and the one or two people holding the set-up together.
The free Who Knows? check covers this in twelve questions, and Under the Bonnet shows the moving parts you now own.
What the team may not see.
Growth targets, cash flow, competitors and the people you answer to all land on your desk first. Most teams never see that side of the business, so the reason behind a push for speed doesn't always travel with the instruction.
What you may not see.
Every piece of AI work needs somebody to brief it, read it, correct it and put their name to it. That work rarely shows in the results. It can look like a slower team when it is actually a team stopping mistakes reaching customers.
The car, the driver and the track.
AI works best when three things suit each other. When something goes wrong, it is worth checking all three.
Your AI tools
Powerful, quick and improving every few months. The part everyone looks at first.
Your people
Their judgement, their knowledge of your customers, and their memory of what has been tried before.
Your business
The processes, sign-offs, data and habits the AI runs on. A good car on a poor track spends a lot of time being repaired.
The rules still apply.
In the UK, using personal data with AI falls under UK GDPR, and the Information Commissioner's Office publishes guidance on AI and data protection. If your sector has its own regulator, its rules on what you say to customers apply to AI-written words exactly as they do to anyone else's.
- Know what personal data goes into which AI tool, and whether the provider uses it to train its models.
- Use business accounts with a data processing agreement, rather than personal ones.
- For higher-risk uses, such as anything involving staff or customer records, do a data protection impact assessment (DPIA) first.
- Update your privacy notice if AI tools now handle customers' or staff details, and know where in the world that data is stored.
- Don't let AI make decisions about people (hiring, pay, credit) without a person deciding.
- Keep regulated or legal wording under a named person's sign-off.
- Keep a simple register of every AI tool: who owns it, what it costs, what data it sees.
This is general information, not legal advice. For your own obligations, check with your data protection lead or adviser. UK data protection law is being updated under the Data (Use and Access) Act 2025, and the ICO's guidance is changing with it.
Worth asking at your next leadership meeting.
If the person who set up our AI processes left on Friday, who could explain them?
Where are our AI tools' instructions written down?
How many hours a week does the team spend checking AI output?
When AI gets something wrong, who owns it?
Who signs off AI work before a customer sees it?
If AI prices doubled next year, who decides what we keep?
Your competitors can buy the same AI.
Four companies ask four AI tools for a campaign idea and get four sensible, interchangeable answers. What makes yours different is what the tools can't supply: your people's feel for your customers and their memory of what worked last time.
Behind the Output
A briefing for directors and owners: a one-page summary, the four places a business can be with AI, seven patterns seen from both sides, and six questions for the board, with what good answers sound like.