AI Chatbot for Restaurant Websites: A Practical Guide

How an AI chatbot for a restaurant website answers menu, hours, and policy questions before customers ever open WhatsApp — and why it complements ordering bots instead of competing with them.

Search “AI chatbot for restaurants” and almost everything you find is about WhatsApp ordering bots. That makes sense — WhatsApp is where Indian restaurants take orders, and it works well for that. KFC India's WhatsApp bot processed over 115,000 orders in its first six months. Nobody is going to out-order WhatsApp on a website widget.

But ordering is not the only conversation a restaurant has with a customer. Long before someone opens WhatsApp to order, they are on your website or your Google listing, deciding whether to come at all — and that conversation happens in a completely different place.

The questions that happen before the order

“Do you have anything vegan?” “Is there parking nearby?” “Are you open on Diwali?” “Can we book the back room for 20 people?” “Do you allow outside cake?” None of these are order questions. They are decision questions — the ones that determine whether someone shows up at all — and right now they get answered by whoever happens to be free to check Instagram DMs or pick up the phone during service.

This is exactly the gap a document-based chatbot fills. It sits on your website, answers from your actual menu, hours, and policies, and never touches ordering or payments. It is not competing with your WhatsApp bot. It is answering the questions that happen before WhatsApp ever opens.

What to upload: four documents cover almost everything

You do not need a large knowledge base to handle most pre-visit questions. Four focused documents cover the vast majority of what people ask a restaurant:

Your menu, with ingredients and allergen notes — not just dish names. “Contains peanuts” answers a real question; “Thai Curry — ₹320” alone does not.
Hours and holiday exceptions, written out explicitly by date where you know them, not just “closed on major holidays.”
House policies — parking, outside food, corkage, dress code, whether you take walk-ins or reservations only, seating time limits on weekends.
Events and private bookings — capacity, minimum spend, how far in advance to ask, and who to contact to actually confirm one.

Write each one the way a customer would ask it, not the way a menu designer would format it. Uploading and embedding takes about five minutes once the documents are ready — the harder part is writing them well, not installing the widget.

A chatbot that only answers from real menu and policy documents will correctly say “I don't have that information” rather than guess at a dish that does not exist or an hours change it does not know about. For a restaurant, an honest “I don't know” is far better than a confident wrong answer about whether a dish contains shellfish.

Where a website chatbot fits next to WhatsApp

Think of it as two channels for two moments. WhatsApp handles the moment someone has already decided to order or book and wants speed. The website widget handles the moment before that — the browsing, comparing, and deciding — where the visitor is not ready to message anyone yet, just looking for a fast answer. Restaurants using chatbots across both moments report roughly 30% faster response times than relying on staff alone, and the technology is projected to save the restaurant industry billions annually in reduced repeat-question overhead by the end of this year.

You do not need to choose one. A widget on your website catches the pre-visit questions; WhatsApp keeps handling the orders it already handles well. What you should avoid is asking one system to do both — a document-based chatbot should never be the thing taking payment details or confirming a live table for tonight. Keep it to what it is good at: answering from what you have already written down.

Where visitors actually see it

Most of your pre-visit traffic is not typing your domain name from memory. It is arriving from a Google search, a Zomato or Swiggy listing, or an Instagram bio link — and it lands on your website only after all of that. Placing the widget on your homepage and menu page catches that visitor at the exact moment they are deciding whether tonight is the night, without asking them to open a separate app or start a WhatsApp thread just to ask if you have a table for six.

It is also worth linking the same website from your Google Business profile and social bios, since that is usually the first place someone looks before they search for your menu directly. The widget only needs to exist in one place — your site — as long as people can actually get there.

The same pattern works beyond restaurants

Salons, gyms, clinics, and local service businesses field an almost identical set of pre-visit questions — pricing, timing, what is included, what to bring, cancellation policy. A salon gets “do you do keratin treatments” and “how long does a full head of highlights take” the same way a restaurant gets “is the rooftop pet-friendly.” If your business already has this information written somewhere, the setup is the same: write it clearly, upload it, and put the widget where people are already looking before they call.

If you are running a product-based storefront instead of a physical location, the same document-first approach applies — just with product specs and shipping policy in place of a menu and table policy.

Start free with your menu and FAQ, and see how many of tonight's DMs it would have already answered. The free plan covers one chatbot and 100 queries a month — enough to test it against a real week of questions before you commit to anything.