AI chatbot for events: what it should answer and how to choose one

An AI chatbot for events, or virtual assistant for events, is a conversational service that answers attendees on behalf of the organiser, before, during and after the event, using the event’s information and, where it has it, each person’s own details. What separates one that repeats FAQs from one that genuinely takes work off your team comes down to three things: the context it works with, the channel it reaches people through and what it does when it doesn’t know the answer.

If you’re looking for one, you probably know the week before the event well: the team’s inbox and WhatsApp full of the same question about the car park, answered by hand again and again. That is the work a good event chatbot should absorb.

What is an AI chatbot for events and how is it different from the FAQ section on your website?

An AI chatbot for events understands the question the way the attendee writes it, in their own words and their own language, and answers with the information the organiser has given it. An FAQ page, by contrast, makes the person hunt for their case among the ones you anticipated.

Between the two sits a third type, the menu chatbot: buttons such as “Programme”, “Getting there”, “Contact”. It’s little more than an FAQ page split across screens. It works as long as the question fits a button, and at an event it almost never does: “I’m arriving on the 9am high-speed train and leaving before dinner, will I make the afternoon session?” doesn’t fit in any menu.

You’ll see the same service under several names: chatbot, AI chatbot, virtual assistant for events. For your decision, the name matters little. What matters is what information it holds, which channel it answers through and what it can do with that information.

What should an AI chatbot for events be able to answer?

An event chatbot has to cover three layers of information and recognise a fourth situation: the question it can’t answer.

1. The event’s information. This is the layer every supplier shows in the demo: programme, speakers, access, parking, catering, transport. “What time does the keynote start?”, “is there a vegan option?”, “where can I park?”. It’s essential, but it’s the easy part.

2. Each attendee’s information. “Does my ticket include the dinner?”, “which hotel am I in?”, “what time does my coach transfer leave?”. The right answer depends on who is asking, and the chatbot can only give it if the organiser holds that data (ticket type, registered sessions, flight, hotel) and has passed it on.

Marta: hi, what time am I being picked up from the hotel tomorrow? 🚌 Chatbot: Good morning, Marta. Your transfer leaves at 8:15 from the main entrance of the Hotel Mirador. You’ll arrive at the venue at 8:45.

(example conversation: names and times are fictional)

A chatbot that only knew the event would have replied with the general timetable for every transfer. Marta would have had to find hers.

3. Changes. Room B moves to Room A, the 12:00 session is pushed back half an hour, the return coach leaves from a different door. The chatbot has to answer with the new version as soon as your team updates it and, better still, alert the people affected before they ask.

4. What it doesn’t know. Someone asks whether they can transfer their ticket to a colleague, or mentions an allergy that isn’t covered by the menu. The useful response here is to recognise that information is missing and pass the enquiry to someone on the team, with the conversation in front of them. Making up a plausible answer is exactly what it must not do.

The first three layers are fairly easy to compare between suppliers. The fourth is the one you see least in a demo, and the one that weighs most on the day.

Six criteria for choosing an AI chatbot for events

To choose an AI chatbot for events, compare six things: the channel, the context for each attendee, languages, handover to the team, control over what it can answer and do, and the data you keep. Use this table as your script for the demo.

Criterion	What it means	What to check in the demo
Channel	Where the attendee writes to it	That you can try it on the channel your audience will use, without installing anything
Context per attendee	Whether the answer changes depending on who asks	Two different profiles asking the same question about their ticket or their travel
Languages	Whether it answers in each person’s language	A question in another language, without you having prepared a translated version
Handover to the team	What happens when it doesn’t know	That an unanswered question reaches someone on your team and that person replies in the same chat
Control	What it can answer and what it can do	A request outside what you have approved
Data	What you keep once the event is over	What you can see of the conversations and how you would use it for the next edition

Channel: where your attendees are

Your audience already has WhatsApp on their phone; an event app, often not. At the events we work on in Europe, WhatsApp tends to be the most natural route, because attendees already have it on their phone and don’t need to download anything. In markets such as the United States, the same conversation can run over SMS and multimedia messaging, depending on how the project is configured. And a chat on the event website serves people who are looking at the programme before they register.

If you’re weighing up whether your event needs its own app or whether WhatsApp is enough, we cover it in what changes between an event app and a WhatsApp assistant.

Context per attendee

Ask in the demo: does it answer everyone the same way, or according to each person’s ticket, sessions and travel? If the supplier only shows you general questions, ask them to answer two different profiles.

For an agency this criterion has a second reading. Cristina, for example, runs an agency with several accounts and doesn’t want a generic chatbot: she wants an assistant with the information for each event and each client, in their tone and with their terms. These custom bots for corporate events are set up event by event, so it’s worth asking how much work it takes your team to build the next one.

Languages

At an international congress, attendees write in their own language and expect the reply in that same language. A good event chatbot does this from the information you already have, without you preparing a version of the content for each language. Check it with a question written in a language you haven’t mentioned to the supplier.

Handover to the team

What happens when it doesn’t know is probably the criterion that most separates one supplier from another. The questions the chatbot can’t resolve have to reach your team, and your team has to be able to pick up that conversation without asking the attendee to start again on another channel.

Control over what it can answer and do

The chatbot should answer with information you have approved and respect what each user is allowed to see: the timing of a private dinner isn’t for everyone. And explaining a policy is not the same as granting an exception: it can set out the conditions for changing a ticket, but authorising a change after the deadline is your team’s decision.

What can go wrong here and how to check it with your supplier is covered in the risks of putting an AI agent in charge of your attendees’ enquiries.

Data you keep

When the event ends, the conversations are the closest thing you have to a record of what your audience wanted to know. The most repeated questions tell you what was missing from the event communications; feedback gathered in the same conversation, while the session was still fresh, tells you what worked. Ask what you get to see of all that and whether you can use it to prepare the next edition.

How to prepare the demo so you choose with your own cases

If you let the supplier choose the questions, you’ll see their best cases, not yours. To compare properly, bring your own material:

• Between five and ten real questions from your last event, copied exactly as they arrived by email or WhatsApp, typos and all.
• One question whose answer isn’t in any document you’ve given them.
• Two different attendee profiles, for example one with dinner included and one without.
• A schedule change made during the demo, to see when it starts answering with the new information.

Run the same tests with every supplier you’re considering. That way you’re comparing answers to the same questions.

How we do it at Sinapser with EVA

EVA (Event Virtual Assistant) is our virtual assistant for events. It looks after your attendees on WhatsApp, with no event app to download, over SMS and multimedia messaging depending on the project configuration, or in a chat built into your website. It answers instantly, 24/7 and in the attendee’s language about the programme, speakers, access or parking, and takes into account their ticket type, their sessions, their travel or their needs, depending on the data available.

The information comes in by connecting your tools or adding it by hand, and your registration and programme can stay in the systems you already use. Your team schedules reminders and last-minute changes for everyone or by segment. When an enquiry needs a person, EVA hands it over and your team carries on the conversation from the dashboard. And what your audience asks and values in those conversations isn’t lost: with EVA Insights you discover their interests, build audiences and put that knowledge to work at your next events.

You can see how it works on the EVA page and the detail of personalised support for each attendee. If you’d rather tell us about your event, write to us at projects@sinapser.com.

Common mistakes

Choosing based on how well it chats in the demo

With easy questions, almost every chatbot looks good. The difference shows with the question that isn’t in the documents and the one that depends on who is asking. If the demo includes neither, you haven’t yet seen what you need to.

Loading only the FAQ page

It’s the quickest thing to prepare and it leaves out what people ask most: each person’s hotel, their transfer, whether their ticket includes dinner. The chatbot ends up answering what was already on the website, and the personal questions keep landing in the team’s inbox.

Not deciding who handles what the chatbot hands over

Picture Thursday at 4pm, in the middle of the exhibition: the chatbot has passed three enquiries to the team and nobody knows they’re theirs. For those attendees, the chatbot has stopped helping: they asked and nobody is answering. Handover needs a name and a rota, just like the registration desk.

Making people download another app to ask something simple

Someone who just wants to know where to park isn’t going to install an app to ask. If the chatbot lives inside an app your audience hasn’t downloaded, that question will end up on the organiser’s phone.

Not updating it when something changes during the event

The room change is announced over the PA and by email, but nobody updates it in the chatbot. From that moment on, every answer about that session sends people to the wrong room. Build the chatbot update into the same protocol you use to announce a change.

Not thinking about what data you want to keep

If you don’t decide before the event what you want to learn (which questions keep coming up, which sessions draw interest, how each part is rated), when it’s over you’ll have plenty of conversations and no conclusions.

Frequently asked questions

Is an AI chatbot for events the same as a virtual assistant for events?

Yes, in practice the terms are used interchangeably. Some suppliers call the button-based one a chatbot and the one that understands what you write in your own words an assistant: ask which one they’re showing you.

Do I need an app to have a chatbot at my event?

No. An event chatbot can work over WhatsApp, SMS or a chat on your website, so attendees ask from their phone without downloading anything. If you also have an event app, the chatbot can sit alongside it.

Which languages does it answer in?

It depends on the supplier, so ask and test it in the demo with a language you haven’t flagged in advance. What you should insist on is that it answers in each attendee’s language without you having to prepare a version of the information for each language.

What happens if the chatbot doesn’t know the answer?

A good chatbot recognises that it’s missing information and passes the enquiry to the organising team, who can carry on the conversation with the attendee. For that to work, someone on your team has to be assigned to handle those enquiries during the event.

Can each event or each client have its own chatbot?

Yes, and for an agency or a company with many corporate events it’s what makes most sense: each chatbot is set up with the information, terms and tone of that event or that client. Ask the supplier how much work it takes to build a new one starting from the previous one.

Is it also useful before and after the event?

Yes. Beforehand, it resolves questions about access, the programme and logistics and sends reminders. During the event, it alerts people to changes. Afterwards, it can ask for feedback in the same conversation and leaves you a record of what your audience asked, useful for preparing the next edition.