How to set up a booking bot for a venue

Setting up a booking bot comes down to five steps: connect a channel, describe the places (tables or specialists), upload the menu, fill in the venue profile, and invite staff with the right roles. The bot answers only what it has been told, so most of the time goes not on the technology but on preparing the information. Below is the order of actions and a list of what to gather in advance.

What to prepare before launch

Don't open the dashboard with an empty head. Gather in one document:

• the venue's name and a short description: what makes it different, who it's for;
• a list of tables, rooms and zones (or specialists and services for a salon) with capacity;
• opening hours and special days;
• the menu in any convenient form: PDF, photos, Excel, Word or just text;
• deposit or minimum spend terms, if you have them;
• the rules for cancelling and moving a booking;
• the dress code and age restrictions;
• the offers currently running;
• answers to the questions guests ask every week (parking, children, pets, the terrace);
• the list of staff and who is responsible for what.

It takes an evening, but afterwards you'll thank yourself: every empty line can turn into a guest's question that the bot can't answer and will call a person about.

Step 1. Channels

Guests write where it's convenient for them, so connect the channel you actually communicate in. GuestKeepr works with Telegram and WhatsApp (via the Cloud API). The dashboard has a Channels section for this.

What to know before connecting:

• Language. The bot answers in the guest's language: Russian, English, Romanian or Armenian.
• Voice messages. The bot turns them into text. For this, a speech recognition key has to be connected on the platform side.
• The WhatsApp window. You can reply to a guest within 24 hours of their message. It's the messenger's rule, and it affects how you build communication.
• The phone number in Telegram. The bot asks the guest to confirm their number with the “share phone number” button: that's how it makes sure the booking belongs to that person.

We wrote more about how the channels differ in the article about bookings via WhatsApp and Telegram.

Step 2. Places: tables or specialists

Everything rests on “resources”: for a club or a restaurant these are tables and VIP rooms, for a beauty salon they're specialists. In the Settings section you set up places and zones. The terms in the interface adjust to the type of venue: “booking” or “appointment”, “table” or “specialist”.

Tips for filling it in:

1. Name tables the way the staff call them on the floor: “Window 3”, “VIP-1”, not abstract numbers nobody remembers.
2. State the capacity honestly. If a table seats four, don't write “up to six” for convenience.
3. Mark out zones: the terrace, the main room, the bar. Guests often ask about them.
4. For a salon, list the specialists, and the bot will be able to show who's free.

The good news about reliability: a double booking of one table for overlapping times is impossible in the system; it's controlled at the database level. A booking whose guests are already seated can't be cancelled or moved.

Step 3. The menu

The bot knows the menu if you've uploaded it. PDF, photos, Excel, Word and plain text are supported. There's also a menu editor in the Settings section. We've put the finer points of layout into a separate article: how to prepare the menu for the bot.

In brief:

• check that the prices and items are current, because the bot will quote them to the guest;
• if you have a seasonal menu, update the file along with the dish cards;
• a tidy text file is better to upload than a blurry photo.

Step 4. The venue profile

This is the bot's main document. In the venue profile the owner fills in:

• the venue description;
• the deposit or minimum spend terms;
• the cancellation terms;
• offers;
• ready answers to frequent questions;
• the dress code;
• age restrictions.

One important clarification. The bot only states the deposit and cancellation terms to the guest as text. It doesn't take payment, doesn't calculate the deposit and doesn't take prepayments. If you need to take money, that will have to be organised separately, outside the bot.

See the article about a booking cancellation policy for how to write terms people understand. The wording should be short and unambiguous: “The deposit is refunded if you cancel at least a day in advance” reads better than a long paragraph.

An example of a ready answer: “Our dress code is smart casual; sportswear and flip-flops aren't allowed. Entry from age 21.” If your rules are different, write your own: this is just an example of the structure.

Step 5. Staff and roles

When the bot is ready, add people. Permissions are set in the Staff and Roles sections, and invitations are sent by email. Signing in with Google is possible, and on the shared tablets on the floor a PIN sign-in is convenient. This is described in detail in the article about staff roles in the dashboard.

Agree in advance:

• who gets the notifications when the bot calls a manager (they arrive in Telegram);
• who on the shift takes over the chat and answers the guest from the dashboard;
• who has the right to cancel and move bookings.

Testing before launch

Before telling guests you have a bot, go through the scenarios as if you were a guest:

1. Ask about free places for the coming Friday.
2. Book a table, move the booking and cancel it.
3. Ask for the menu.
4. Ask about the dress code, the deposit and cancellation.
5. Ask a question the bot shouldn't know the answer to, and make sure it calls a person.
6. Write in English, and send a voice message.

If at some step the answer is wrong or empty, an entry is missing from the profile: add it and repeat.

We don't give prices or dates: they haven't been set yet.

After launch

In the first week, read the conversations in the Conversations section and look at where the bot called a manager. Each such case shows what's missing from the profile. In a couple of weeks the set of ready answers will be noticeably fuller. How to divide work between the bot and staff is covered in the article about AI and human front desks.

To sum up the approach briefly: the GuestKeepr virtual front desk works as well as you've described your venue.

Typical setup mistakes

• The menu was copied from the website along with the old prices. Check every price against what's on the floor today.
• Rules that are too long. The guest reads in a messenger on a phone. Nobody will finish a ten-line paragraph, so split the terms into short points.
• No one was put in charge. When the bot calls a manager, the notification must reach a specific person on the shift, not “everyone and no one”.
• The rescheduling scenario wasn't tested. Cancelling and moving most often break on details: a different time, a different table, a different number of guests. Go through them by hand.
• Launched and forgotten. The venue profile lives along with the venue: a new menu, a new offer, new opening hours all need editing.

Frequently asked questions