Before you start8 steps · about 8 minutes
Before you start
- A list of the AI tools you and your team already use
- A rough idea of what you use them for
- Your privacy policy, as you may need to update it
- About 20 minutes
Good to know: What AI is: a tool, an assistant and a productivity boost. What it isn't: always correct, something to trust blindly, or a replacement for you.
Step 1 of 8
1Decide what AI can and can't do
Write a short list. Jobs AI can help with: drafts, summaries, ideas. Jobs that stay human: pricing, advice to customers, final approvals and anything legal or financial.
Step 2 of 8
2Always check the facts
AI can give confident answers that are simply wrong, often called hallucinations. Check every fact, figure, price, date and link before anything goes public.
Step 3 of 8
3Keep customer data out
Don't paste customer names, contact details or anything confidential into a personal AI account. Google itself asks Gemini users not to enter confidential information they wouldn't want a reviewer to see.
Step 4 of 8
4Check your settings
On personal ChatGPT accounts, chats can be used to train models unless you turn off "Improve the model for everyone" in Settings, then Data controls. ChatGPT Business and Enterprise don't use your content for training by default.
Step 5 of 8
5Keep your voice
Edit every AI draft until it sounds like you, and add your real stories, jobs and opinions. That's the part competitors can't copy.
Step 6 of 8
6Add value, don't just add pages
Google is fine with AI-assisted content that's genuinely helpful. Churning out lots of pages without adding value can break its spam policy on scaled content abuse.
Step 7 of 8
7Be open where it matters
If a customer is chatting with a bot, say so and make it easy to reach a person. Never use AI to write reviews or testimonials.
Step 8 of 8
8Review it regularly
Every few months, check what your team uses AI for, what's working and whether anything needs tightening up.
All 8 steps done
Guide complete. Nice work.
Human in the loop: you. Nicely done. Keep a copy of the whole guide, with every step, tip and source, for next time.
Stuck on anything? Tell us where and we'll pick it up from there.
Why AI gets things wrong
AI chat tools predict the most likely-sounding answer. OpenAI's own research explains that the way models are trained and scored rewards guessing over saying "I don't know". That's why AI makes a brilliant assistant and a risky decision-maker. Your knowledge, judgement and relationships are what customers are really paying for.
Best-practice tips
- Treat it like a new starterBrilliant at first drafts, needs supervising, never signs anything off.
- Name a human sign-offOne named person approves anything AI-assisted before customers see it.
- Use business accountsIf your team uses AI a lot, a business plan gives you admin controls and stronger data protection.
- Write it downA one-page AI policy for your team beats everyone guessing.
- Stay human with customersUse AI behind the scenes and keep your customer conversations personal.
- Keep learningTools change fast. Our free monthly workshops at the hive cover what's new.
Questions, answered
Will AI replace my team?
So far it's much better at speeding up tasks than replacing people. The ONS found that rising AI use hasn't yet led to widespread changes in headcount at UK businesses.
Does Google penalise AI-written content?
Not for being AI-written. Google cares whether content is helpful and accurate. Mass-producing low-value pages is what breaks its rules.
Is it safe to put customer data into ChatGPT?
Not into a personal account. If you need AI to work with personal data, use a business plan, check the provider's terms and your UK GDPR obligations, and mention it in your privacy policy.
Should I tell customers I use AI?
Be open wherever it affects them directly, such as a chatbot or AI-generated images. For behind-the-scenes drafts that you check and edit yourself, labelling every email isn't necessary.
- Data Controls FAQ (OpenAI Help) (opens in a new tab)
- Gemini Apps Privacy Hub (Google) (opens in a new tab)
- Google Search's guidance on generative AI content (Google Search Central) (opens in a new tab)
- Why language models hallucinate (OpenAI) (opens in a new tab)
- Artificial intelligence and data protection (ICO) (opens in a new tab)
- Artificial intelligence in UK businesses: 2023 to 2026 (ONS) (opens in a new tab)