Should our website tell customers when an AI tool is recommending products instead of a person?
Yes. Make it clear that the recommendations are generated by AI, explain the information used to produce them, and give customers an easy way to reach a person before they act. Do not present a generated recommendation as advice from an employee who has not reviewed it.
Ask a forward-deployed engineer to build the recommendation experience around verified product information and clear limits. The screen should distinguish a suggested product from confirmed availability, pricing, or suitability. If the tool lacks enough information, it should say so rather than fill the gap with a confident guess.
A practical design might include:
- A visible notice near the recommendation, not only in a privacy policy.
- Links to the product details supporting the suggestion.
- A way to correct an input and see the recommendation change.
- An option to contact a knowledgeable employee without starting over.
Review the experience with the people responsible for product information and customer relationships. Decide who updates the source material and who handles questions the tool cannot answer. Check applicable disclosure and privacy requirements, particularly if the recommendation could have significant consequences for the customer.
The EOS Model encourages clarity and accountability. Applied here, that means customers understand what the tool is doing, while a named person owns its accuracy and escalation path.
Category: AI that stays human