Supplement Sellers: Answering Product Questions Safely (No Medical Claims)

A supplement seller on WhatsApp in Malaysia should let an AI agent share only approved product facts, such as ingredients, how to take it as printed on the label, price and delivery, and never answer 'will this cure my problem?'. Any question about a condition, a medicine or a pregnancy goes to a pharmacist or doctor, not to a chatbot or a salesperson.

Why is answering supplement questions on WhatsApp risky?

Answering supplement questions on WhatsApp is risky because customers ask health questions in the same breath as price questions, and a careless reply becomes a health claim in writing. A screenshot of that chat can travel further than any advertisement.

The typical messages are: 'Can I take this with my blood pressure medicine?', 'My mother has diabetes, can it help?', 'Does this make my skin white?'. A seller who wants to be helpful and close the sale is tempted to say yes. That is exactly the reply that creates legal and safety risk.

This guide does not give legal advice. It shows a practical set of rules for how an AI agent should reply, so the sales process works without the AI acting as a doctor. Have your own legal adviser or regulatory consultant check the final wording. Our earlier post on AI chatbots in the supplement industry gives the general picture; this page is about the reply rules.

What does Malaysian advertising law say about health products?

Malaysian rules restrict advertising of medicines and medical claims, and what you write in a chat can count as advertising in practice. Check the primary sources, not hearsay.

• The Medicines (Advertisement and Sale) Act 1956 and its regulations are published by the Pharmaceutical Services Programme.
• The Ministry of Health explains whether internet medical advertisements need MAB approval.
• The Medicine Advertisements Board publishes advertising guidelines for healthcare facilities and services.
• KPDN enforces the rules on trade descriptions, including penalties for misleading descriptions.

You do not have to memorise these. You need a rule sheet that your team and your AI follow, written with advice from someone who knows your product category.

Which questions can the AI answer, and which must it redirect?

The AI can answer questions about facts on the label and the order, and must redirect any question that asks about a person's health. The table sorts the common ones.

What does a safe conversation look like?

Illustrative example.

The AI stayed warm and useful. It gave facts, avoided any promise and sent the health decision to a professional. That is the model for every health-related question.

How do you build the approved and banned wording lists?

You build them from the product label, the registration documents and your regulatory adviser's advice, then keep each list on one page. Short lists get followed.

1. Approved facts: name, form, pack size, ingredient list, directions as printed, storage, expiry practice, price, packages.
2. Approved comparisons: only those you can prove, such as pack size or price per capsule.
3. Banned verbs: cure, treat, heal, prevent, reverse, detox, lose weight fast, guaranteed.
4. Banned stories: customer testimonials about health results, before-and-after claims, doctor endorsements you cannot document.
5. Redirect lines: the standard sentences the AI uses to send people to a professional.

Give resellers and staff the same sheet. If an AI is careful but a human admin writes 'this one very good for diabetes' on the same number, the risk is unchanged. Our post on a WhatsApp AI assistant for staff SOP shows how to keep a team on one script.

How should the AI handle orders, repeat purchases and delivery?

The AI should handle orders, packages, delivery and repeat-purchase reminders like any other product, because these are commercial questions with no health content. This is where it saves the most time.

• Orders: item, quantity, address and payment, with the order summary confirmed back.
• Packages: explain bundles by price and quantity, not by results.
• Repeat reminders: 'Your pack of 30 may be running low. Would you like to reorder?' is a stock and timing message, not a health message, as long as you do not say it is time to continue treatment.
• Delivery and COD: postage zones, courier and payment proof, as in our guide to payment proof and COD.

Marketing messages need opt-in. See our posts on the WhatsApp opt-in list and on whether WhatsApp blasting is legal in Malaysia.

What should you do with health information customers volunteer?

You should avoid collecting it, because information about a person's physical or mental health is sensitive personal data under the PDPA. The AI should not ask for medical details, and should not store them in the CRM.

If a customer volunteers a condition, the AI gives the redirect line and tags the chat only as 'health question, redirected', not with the condition. The PDP Department FAQ explains sensitive personal data. Ask your adviser how this applies to your records.

How should halal and registration questions be answered?

The AI should state halal status only when you hold a certificate, and give the issuer and details as documented. Customers can verify on the Halal Malaysia Portal.

If you do not have documents, the answer is 'We do not hold halal certification for this product'. Do not hint, imply or say 'all ingredients are plant-based, so should be okay'. The same discipline applies to product registration: show what you can document and nothing else.

When should a human step in?

A human should step in when a customer reports a reaction, an adverse effect, a complaint about the product, or anything that sounds like an emergency. The AI should stop selling and give a safe response.

• Reports of side effects: staff respond, record the case and follow your product-safety process.
• Emergency words: the AI says to contact a doctor or call emergency services, and tells staff.
• Press, authority or legal enquiries: straight to the owner.

How does a ChatsHero AI Agent Office keep a supplement shop compliant?

A ChatsHero AI Agent Office gives a supplement shop a receptionist and sales employee that work only from your approved sheet, a redirect rule for health questions, and a human hand-over for complaints and side-effect reports. ChatsHero, a Malaysian AI agent platform, builds these from your own SOP, price list and FAQ.

The AI does not replace a regulatory adviser. It applies the rules you give it, consistently, at every hour, and your weekly chat review catches what slips through. A free 30-minute Zoom demo using your own most-asked questions will show you how it responds. Read the overview in what an AI Agent Office is.

Frequently Asked Questions