An AI Chatbot on Your Website: When It Helps and When It Hurts

The promise is real

Language models have made chatbots far more useful than the old "press 1 for sales" scripts. A well-built assistant can answer questions about your services at any hour, help a visitor find the right product, collect the details of a request and pass a warm lead to your team in the morning.

But a chatbot speaks on behalf of your company. Everything it says, your customers will treat as your word.

A lesson from Air Canada

In February 2024, British Columbia's Civil Resolution Tribunal ruled in Moffatt v. Air Canada. The airline's website chatbot told a customer he could apply for a bereavement fare discount after travelling — which contradicted the airline's actual policy. Air Canada argued that the chatbot was responsible for its own actions. The tribunal rejected that argument: the chatbot is part of the company's website, and the company is responsible for the information on it. Air Canada was ordered to compensate the customer.

The amount was small; the principle is not. If your bot invents a price, a guarantee or a policy, you may have to stand behind it.

When a chatbot makes sense

• You receive many repetitive questions: prices, delivery, opening hours, how to book
• Your clients write outside working hours or in several languages
• You have good, up-to-date content the bot can rely on — FAQ, service pages, documentation

If you get five enquiries a week, a clear contact form and a WhatsApp button may serve you better.

How to build one that doesn't embarrass you

• Answer from your own content. Connect the model to your approved texts and instruct it to say "I don't know" rather than guess
• Keep sensitive topics human. Prices for custom work, contracts, refunds and complaints go to a person
• Make the hand-off easy. A visible "talk to a human" option, with the conversation passed along so the client doesn't repeat themselves
• Be transparent. Tell visitors they are talking to an AI assistant
• Protect data. Don't ask for more personal information than needed, and check where conversations are stored and processed
• Read the logs. Review conversations regularly; they show both the bot's mistakes and the questions your website fails to answer

Measure the result

Decide in advance what success looks like: fewer routine emails, more booked calls, faster first response. Without a metric, a chatbot easily becomes a widget that slows the page and talks to nobody.

A chatbot works best next to a good form, not instead of it — see how to design lead forms that convert. To build an assistant yourself, start with the Academy guide to building an AI assistant or the AI sales manager course; to have it built and integrated for you, see AI automation services.

Considering an AI assistant for your site? Codearia designs and integrates chatbots that work from your own content and hand off to your team — tell us what you want to automate.