The opening eBook of the AI, Automation and Tools category. It explains, in plain language, what AI can and can't do for a small business with one to twenty people, and gives you a working set of habits you can adopt this week without spending much money or learning to code.
Members ebook·7 chapters· 40 minute read
Chapter 5
AI for Customer Service and Sales Conversations
Where AI helps you talk to customers - and where it should stay well out of the way.
Talking to customers is the part of a small business where the gap between good and bad is widest. A warm, well-judged reply turns a tricky moment into a story the customer tells their friends. A clumsy auto-reply turns a small problem into a lost customer and a one-star review. AI sits awkwardly in the middle of this. It can help you reply faster and better. It can also do real damage if you let it talk to customers in your name without supervision.
The right way to use AI here is as a backstage assistant, not a front-of-house one. It drafts. You send. It summarises. You decide. It prepares. You speak. The customer should never know the assistant exists. That sounds restrictive. In practice, it's where most of the value is.
This chapter sets out where AI helps in customer service and sales, and the lines you shouldn't cross.
The full chapter covers reply drafting, sales call preparation, follow-up sequences and the small print on chatbots, with the simple rule that decides which jobs to hand over and which to keep.
Drafting replies, not sending them
The single most useful customer-facing job for AI is drafting replies to incoming messages. A complaint email. A request for a complicated quote. A polite but awkward question about a refund. A long message from a long-standing customer that needs to be answered carefully.
The pattern is the same every time. Paste the customer's message into the assistant. Tell it the situation in one or two lines - what you're willing to offer, what you're not, anything you want to acknowledge. Ask for a draft reply in your voice. Read what you get. Edit the bits that don't sound like you. Send the edited version. The whole job, from received to sent, is often under ten minutes for a message that would have taken thirty.
What you've done is keep your judgment in the loop. The customer reads a reply you sent. The assistant just made the typing faster.
Preparing for sales calls and meetings
Before a sales call or a first meeting, the assistant is excellent for preparation. Tell it who you're meeting, what you know about them, what you're hoping to achieve and what you're worried about. Ask for a list of questions to ask, the three things they're most likely to be worried about, and the two objections they're most likely to raise.
What you'll get is a workable prep sheet. You'll add things from your own knowledge. You'll cross things out that don't apply. The act of working through the prep with the assistant means you arrive at the call with clearer thinking, not a script. The customer doesn't need to know any of this happened. They just notice that you came prepared.
Three rules for AI in customer conversations
Use it to draft, prepare and summarise. Always read what it produces
Never let it send a message in your name without you reading it first
If a customer would feel deceived to know AI was involved, don't use it for that bit
Follow-up sequences after a quote or a conversation
After a quote goes out, most small businesses follow up too rarely or too aggressively. The assistant is a useful middle ground. Brief it once on your follow-up rhythm - say, a check-in three days after the quote, a friendly nudge a week later, a final reminder two weeks later, all in your voice. It will draft the three messages. You edit them once and save them as templates. Now every future quote has a three-message follow-up sequence ready to go, with two minutes of editing per customer to make it personal.
The same pattern works for renewals, second-purchase nudges, win-back messages and the after-the-event thank-you. The assistant doesn't replace the relationship. It removes the friction that stops you from showing up consistently.
Chatbots on your website - approach with care
It's now possible, and increasingly cheap, to put an AI chatbot on your website that talks to visitors. For most small businesses, this is the wrong move. The cost of a bot that says something wrong - quoting a price you can't honour, promising a service you don't offer, telling a customer something inaccurate about your opening hours - is far higher than the value of replying twenty seconds faster than you would have done.
If you do want a bot, two rules. First, scope it tightly to a small number of safe topics where you've personally written and tested the answers - opening hours, location, how to book, how to get a quote. Second, always offer a clear way for the customer to talk to a real person within a couple of clicks. The bot is there to help, not to keep customers away from you.
Summarising customer conversations and notes
After a long meeting, a long email thread or a difficult phone call, dump your notes or the transcript into the assistant. Ask for a one-paragraph summary, a list of action points and any commitments you made. This takes thirty seconds and prevents the small but expensive mistake of forgetting something you promised. For sensitive conversations - health, legal, anything covered by confidentiality - skip this. Take the notes yourself by hand.
The line you don't cross
There's one line worth holding firmly. If a customer would feel deceived to know AI was involved in a particular message - a sympathy note, an apology after a real mistake, a personal recommendation - write that one yourself. AI is for routine and for speed. The meaningful moments are why people stay loyal. They deserve your time.
What to do this week
Pick the next awkward customer message that lands in your inbox. Draft your reply with the assistant's help. Edit it. Send it. Notice how long it took compared to the last one you wrote unaided. Then pick one quote you sent last week that's gone quiet and use the assistant to draft a friendly follow-up.
Recurring principle for this chapter: use AI and tools to support judgment, not replace it. For more on the underlying customer service philosophy, look back at Customer Service and Experience. For the next step on the back-office time savings, look ahead to chapter six.
The rest of this chapter walks through the practical steps, the templates and the checklists you need to put it into action. It includes worked examples, copy frameworks and the small decisions that make the difference between a plan that sits in a drive and one that gets used.
Inside you'll find a step-by-step playbook, a downloadable template, a checklist you can run this week and a short list of common mistakes to avoid before you start.
The full action plan, broken into weekly steps.
Ready-to-use scripts, templates and checklists.
Worked examples for different sized businesses.
Common mistakes and how to avoid them.
Members-only chapter
Become a member to read the full chapter
Members get the complete chapter, the step-by-step plan, the templates and the checklists. Cancel anytime.
