An AI assistant on your website speaks for your business. Before you let one, from us or anyone else, get clear answers to these seven questions. A good provider will have them ready.
1. Where do its answers come from?
It should answer from information you supplied and approved, not from the open internet. Ask to see the exact text it draws on.
2. What does it do when it doesn’t know?
The right behaviour is to say so and offer to pass the question to a person. Test this yourself with a question that is not covered.
3. What is it not allowed to say?
Final prices, financing terms, diagnoses and promises about timing should stay with you. Check that these limits are written into its setup.
4. Does it tell customers it is an AI?
It should, at the start of every conversation, and it should say if the conversation may be recorded. This is basic honesty, and in some places it is the law.
5. What happens to the conversations?
- How long are they kept?
- Are they used to train AI models?
- What happens if a customer types something sensitive, such as a card number?
6. How do requests reach me?
A request that sits in a dashboard you never open is a lost customer. Ask for it to arrive where you already look, and send a test to make sure.
7. Who fixes it when it says something wrong?
It will happen eventually. You want a named person, a way to report it, and a change made quickly. Ask how conversations are reviewed after launch.
Rules on AI disclosure and call recording vary by place, so check yours; this is not legal advice. If you run a medical, dental, legal or financial practice, add an eighth question about privacy law in your field before going further.
