AI Allergen Chatbot Restaurant: 14 Allergens Covered Safely

A guest who asks whether there are peanuts in the pesto deserves a clear and correct answer, not hesitation from the waiting staff or a guess from memory. AI allergen chatbot restaurant ensures that answer is the same on every channel: on your website, in WhatsApp, in a reservation confirmation or in a conversation with the AI receptionist. A guest with a nut allergy immediately hears that the pesto contains walnuts, while a vegetarian guest looking at the same dish instantly sees whether it is animal-free or not.

According to the Netherlands Food and Consumer Product Safety Authority, correct information about the fourteen legal allergens is mandatory in all hospitality venues, and Allergy Association Netherlands notes that incorrect or missing information is still one of the biggest causes of avoidable allergic reactions in restaurants.

In this guide you will read how an AI allergen chatbot works, how to structure your menu data properly, how to fully cover the fourteen legal allergens and which pitfalls are best avoided.

What is an AI allergen chatbot for restaurants?

An AI allergen chatbot for restaurants is an AI system that knows for each dish which ingredients it contains and which of the fourteen legal allergens are present or may be present. Guests can ask questions in natural language, for example "I have a gluten allergy, which starters can I have", and receive a direct and correct answer.

For waiting staff this means fewer questions during busy service and less doubt about what they are allowed to say. For guests it creates a sense of safety and the certainty that the same information can be found across every channel of the restaurant.

Practical tip: in the first month start with the four most common allergens (gluten, nuts, lactose and shellfish) and only then expand to the other ten. This allows the kitchen to gain confidence in the process without the project plan getting stuck in completeness.

The fourteen legal allergens

The European Food Information Regulation requires restaurants to provide information about these fourteen allergens: gluten, shellfish, eggs, fish, peanuts, soy, milk and lactose, nuts, celery, mustard, sesame seeds, sulphites, lupin and molluscs.

Our AI covers these fourteen as standard, distinguishing between "contains" and "may contain traces" and taking cross-contamination in the kitchen into account. For unknown or regional products we involve a dietitian during implementation to validate the classification.

How do you structure the menu data?

A good allergen chatbot is only as good as the menu data behind it. We typically use a three-layer approach that allows both the kitchen and the front-of-house team to maintain the system quickly:

• Ingredient layer: each ingredient records which allergens it contains, based on purchasing data and product specifications from your supplier.
• Recipe layer: each dish records the ingredients with quantities and preparation method, including sauces and garnishes.
• Kitchen layer: each kitchen records cross-contamination risks, for example a shared fryer or a shared grill.

Also read how a chatbot handles menu questions and allergens and how an AI chatbot for restaurants works completely in 2026.

Which channels does the chatbot cover?

In practice guests ask about allergens in many different places. A good AI covers these five standard channels, all based on the same menu data so answers remain consistent:

• Web chat on your restaurant website for guests considering a booking.
• WhatsApp for guests who have already started a conversation with your restaurant.
• AI receptionist for guests who call directly with a question.
• Reservation confirmation with a proactive question about allergens and dietary requirements.
• At the table via QR code with a digital menu that can be filtered per guest.

GDPR and the European AI Act

An allergen question contains health data and therefore sensitive personal data under GDPR. The rules:

• Under the European AI Act, information affecting guest safety falls into a higher risk category and the source of the answer must be traceable.
• The Dutch Data Protection Authority requires a short retention period for allergy information and prohibits linking it to a marketing profile without explicit consent.
• For a guest with a severe allergy the conversation is always handed over to a staff member for additional verification.

We include this as standard in the implementation, including European hosting of transcripts and a short retention period for allergen-related data.

What the chatbot deliberately does not do

An AI allergen chatbot is not medical advice and not a replacement for the head chef. We deliberately never allow the bot to independently decide on:

• A medical statement about an allergy or intolerance.
• A guarantee that a dish is completely free of a specific allergen when a cross-contamination risk exists in the kitchen.
• A commitment about allergens in a new dish that has not yet been approved by the chef.

In those cases the guest is politely advised to speak with the chef or the host on arrival or before ordering. In practice this is often taken as a compliment rather than something the bot should handle itself, because the guest feels that the restaurant takes their safety seriously.

Which KPIs should you track?

An allergen chatbot is only valuable if you can see what it delivers. For our clients we track five KPIs that together provide insight into quality and safety:

• Coverage percentage of the fourteen legal allergens across all dishes on the menu.
• Number of questions per channel with a distinction between web chat, WhatsApp and phone.
• Number of warm handovers in severe or complex allergy situations.
• Lead time from a menu change to updated allergen information across all channels.
• Sample verification by the chef or a dietitian on a random selection of answers.

Common mistakes during implementation

In the first months we regularly see three mistakes: using outdated menu data, not distinguishing between "contains" and "may contain traces", and failing to arrange a warm handover for serious allergy questions. Our chatbot for menu questions and allergens guide goes deeper into the data structure behind a reliable allergen chatbot. Start small, verify through sampling and only expand based on data.

Get started with HorecaHub

Want to know what an AI allergen chatbot would look like for your restaurant and how many questions your team could save per service? Explore the options on the pricing page or schedule a no‑obligation conversation via contact. During that conversation we demonstrate live how the chatbot gives the same answers on web chat, WhatsApp and phone, which integrations we set up first with your menu system and which measurable results our restaurant clients achieve in the first three months, so you can build a realistic case for your owner or management team.