AI chatbot examples: what a good conversation looks like
A chatbot on a service business site has a narrow job. It answers the questions visitors ask most, collects the request and passes it to a person. The five conversations below show that job in five trades. All of them are illustrative: the businesses are fictional, and we wrote the dialogue to show the pattern, not to report results.
If you are comparing terms first, AI agent vs chatbot vs AI assistant explains the difference. For the practical product, see our AI assistant for small businesses.
Example 1: a plumber’s website chat (service area and quote request)
Illustrative example.
Visitor: Do you work in Somerville on weekends?
Chat: Hi, I’m the AI assistant for this plumbing company. Yes, we serve Somerville. Weekend visits are available for urgent jobs. What do you need help with?
Visitor: Water heater is leaking.
Chat: Sorry to hear that. If water is spreading, turn off the water supply to the heater if you can. May I have your name and the best number to reach you?
Visitor: Sam, 555-0100.
Chat: Thanks, Sam. I’ve sent your request and the details of the leak to our team. Someone will call you back. Would you like a text confirmation?
What to notice. It says it is an AI. It answers from the service area and hours the owner approved. It collects only a name and a number. It does not promise a time, a price or a technician it cannot confirm.
Example 2: a salon’s chat (price range and booking link)
Illustrative example.
Visitor: How much is balayage?
Chat: Balayage starts at a price that depends on hair length and the stylist. Our price list shows the range, and the stylist confirms the exact price at your consultation. Would you like the booking link?
Visitor: Yes, and do you do men’s cuts?
Chat: We do. Here is the link to book a cut or a balayage consultation. If you tell me your preferred day, I’ll note it for the front desk.
What to notice. The owner decided which prices to share and which to leave to a consultation, so the bot gives a range and not a guess. It points to the booking tool you already use.
Example 3: a dental office chat (general answers, no patient information)
Illustrative example.
Visitor: Do you take Acme Dental Plan, and can I get an appointment for a toothache tomorrow?
Chat: I’m an AI assistant. Yes, the office accepts Acme Dental Plan. I can’t take health details or book clinical visits in this chat. Please call the front desk at the number below, or use the practice’s booking page. If you have severe swelling or trouble breathing, call 911.
What to notice. The insurance answer comes from a list the practice approved. The chat declines to collect symptoms or a name tied to an appointment, because under HIPAA that information counts as protected health information (45 CFR 160.103) and may only go through tools whose vendor signs a business associate agreement. The emergency line is a message the practice writes in advance.
Example 4: a law firm chat (intake without legal advice)
Illustrative example.
Visitor: My landlord is keeping my deposit. Can I sue?
Chat: I’m an AI assistant and I can’t give legal advice. I can take a few details so an attorney’s office can review whether it can help: your name, a phone number or email, and a one-line description of the matter. Contacting us does not create an attorney-client relationship.
What to notice. Intake and scheduling only. The bot says what it cannot do, keeps the questions few and does not judge the case. Firms should check their state’s advertising and confidentiality rules. This is not legal advice.
Example 5: an HVAC company after hours (triage and alert)
Illustrative example.
Visitor: No heat and it’s 10pm. Can someone come tonight?
Chat: I’m an AI assistant for this heating company. Is anyone in danger, or do you smell gas?
Visitor: No, just cold.
Chat: Thanks. I’m sending an urgent after-hours request to the on-call team now. What is your name, number and the address area? Someone will contact you according to our after-hours policy.
What to notice. The first question is about safety, with a message the company wrote for gas smells. The bot alerts the on-call tech and does not promise arrival times. Whether a phone call can transfer live to the tech depends on the phone setup and is agreed in writing. See after-hours emergency calls for HVAC and plumbing.
A bad example: what a chatbot should not do
Illustrative example of what to avoid.
Visitor: How much to replace my furnace?
Bot: That will be about $5,000. I’ve booked you for tomorrow at 9am.
The $5,000 is an invented number, not a market price. The bot guessed a price the owner never approved and booked a visit nobody confirmed. Each of those is a rule problem, not a language problem. A chatbot should only quote what you approved and only book what your calendar allows.
Rule-based and generative AI chatbot examples
A rule-based chatbot shows buttons (“Hours,” “Get a quote”) and fixed replies. It is predictable and breaks on unexpected wording. A generative chatbot writes an answer to a free-text question, ideally from a list you approved. The risk is that a loosely set up one guesses. Many business tools mix both: buttons for common paths, a language model for free text, and a handoff when neither fits. Anthropic’s advice for building with language models is to start with the simplest approach that works and add complexity only when needed.
General-purpose assistants like ChatGPT are a different category. They help one person with drafting and research and do not answer your customers on their own. See ChatGPT for small business.
How to judge a chatbot before you choose
Skip star ratings and brand lists. Use a checklist.
- Approved answers. Can you see and edit every answer it is allowed to give?
- Refusals. What does it say when it does not know, or when the question is medical or legal?
- Handoff. Who receives the conversation, how fast and with what summary? See chatbot to human handoff.
- Logs. Can you read every conversation?
- Channels. Website only, or also texts and Instagram? Instagram has its own rules: Meta’s documentation says an app can message a user only after that user has messaged the account first. See Instagram auto reply.
- Install. Can it run on your site? See how to add a chatbot to your website.
- Price. Monthly fee, usage allowance and what happens above it. Market ranges are on the AI chatbot pricing page.
Where Magnet Mint fits
We set up an AI chatbot for your website that answers from replies you approve, collects the request and hands the lead to your team. Website chat and messaging is from $199/month, with setup from $900. Not included in any price: domain and hosting, third-party software subscriptions (billed at cost), SMS carrier registration (A2P 10DLC) and message fees, and AI usage above the plan allowance.
Want to see what a chatbot would say for your business? Ask for a free demo.
Published . Updated .