What organizers settle first for an AI avatar on stage, and what comes last
In short
The invitation goes out before the briefing. It should mention the AI avatar anyway, by name, even while its position is still open. That is why the character is settled before the invitation. What it should argue can wait until the dress rehearsal.
Last updated 4 October 2026
- Guide
Does an AI avatar at an event have to be labeled?
In the EU, yes, since August 2, 2026. Anyone talking to an AI avatar must be told by the first exchange, unless it is obvious. The provider handles that. If the avatar convincingly recreates a real person, the deployer must also disclose that it is artificial. Switzerland has no AI law of its own yet; we label it anyway, at every event.
- Guide
What goes into the briefing for an AI avatar?
The briefing for an AI avatar contains the topic as a contested question, the names of the people on stage, and the position it should argue. For you, that is a few lines. You only change its name if someone on stage has a similar one.
- Guide
How do you moderate a panel discussion with an AI avatar?
Call on the AI avatar by name, with name and question in one sentence. Your question decides how long he answers: “in einem Satz” gets one sentence, “erklär uns” (explain to us) a full explanation of a good half minute. In your preparation, he is the one guest whose position you don’t ask about. You set it in the briefing.
- Guide
What do you do when the AI avatar on stage goes silent or gets it wrong?
If the AI avatar stays silent after being called, the moderator checks the tablet: without the question there, they call it again with name and question; with it, they tap ANSWER. An answer that goes off track gets stopped, and a guest takes the floor. Without internet it cannot speak; a practiced line then passes its question to the panel.
- Guide
Checklist for an AI avatar at your event, from the first request to the photo caption
Four parties share the work: you as the organizer, your sound engineer, the moderator and Aura51. The roughly two weeks of lead time are mostly for your sound engineer; the briefing is only needed by the dress rehearsal. That is where everyone practices their part once. After the event, the photo caption remains.
- Guide
What does an AI avatar on stage record, and where is it kept?
The AI avatar hears what your sound engineer sends from the mixing desk to our laptop, by default only the stage microphones. A cloud service transcribes, separated by speaker. The transcript is stored on our laptop; for composing and checking, the text goes to language models in the cloud. Tell your stage guests in the invitation.
- Guide
Chatbot, video avatar, interactive AI avatar: what is the difference?
A chatbot answers one person, whatever they type. A video avatar speaks a script that was fixed in advance and does not listen. An interactive AI avatar listens and answers live, with a face and a voice. On stage it also has to recognize when it is being addressed, and stay quiet while people talk to each other.
- Guide
Town hall meeting: how the critical questions get asked
A town hall meeting, also called an all-hands, is a meeting of the whole workforce with a Q&A: leadership stands in front of everyone, reports on where things stand and answers questions. Critical questions rarely come on their own, because the people asking are speaking in front of their own bosses. Anonymous questions that others can upvote help, and so does a moderator who follows up on evasive answers.
- Guide
What is an interactive AI avatar?
An AI avatar is a computer-generated face and voice speaking words that a language model writes. A video avatar speaks a script fixed in advance. An interactive AI avatar listens and answers in the moment, because speech recognition transcribes what is said. At events, it welcomes guests, advises visitors at a booth, guides a program or joins a panel.
- Guide
Is a digital human the same as an AI avatar?
Not quite. A digital human, also called a digital person, is a computer-generated likeness that looks like a person, usually photorealistic. The term doesn’t say whether an actor or an AI supplies the words. AI avatar says who is speaking. That is why we call our characters AI avatars. Anyone listening to them should know that an AI is talking.
- Guide
AI moderator or AI participant: who runs the conversation?
An AI moderator, also called a digital or virtual moderator, runs the program and decides who gets the floor. An AI participant joins the discussion and speaks when the moderator calls on it. The Aura51 AI avatars are participants, and we do not offer a virtual moderator. Whoever moderates has to stay neutral and react to the room instantly. Our AI avatars are not built for that.
- Guide
Why doesn’t an AI answer as fast as a person?
Mainly because it first has to detect that the question is over. People plan their answer while the question is still running and start answering a yes-or-no question 208 milliseconds after it ends, on average. An AI waits for silence. In our live tests, the first sound came after 2–3 s, a multiple of the everyday pause.
- Guide
AI avatar glossary for organizers and sound engineers
An AI avatar on stage brings words usually reserved for the sound engineer or the legal team: aux send and pre-fader in the tech rider, deepfake in the EU AI Act. Others, such as conversation window, first audio and evidence level, have a narrow meaning at Aura51. Measured values come from our own tests, live or in the lab.
01Order
Why the invitation comes before the briefing
Invitations to the guests on stage often go out before anyone knows which position the AI avatar will take. They should mention it anyway. If your event takes place in the EU, the AI Act requires that guests and moderator learn they are talking to an AI no later than the start of the conversation (Article 50(1) and (5)); in Switzerland this is generally not binding, and an early notice there is good practice. If the invitation says so, they know before they accept.
The order follows from that.
- 01
Deciding on disclosure
First you decide that the invitation mentions the AI avatar. It is the first thing your guests read about it. On screen, its image carries the label “AI avatar” in any case, at every event.
- 02
Data protection roles
Before the invitation goes out, we put in writing with you who is the controller of the data and who processes it on the controller’s behalf. The invitation has to state it.
- 03
The character
Which of the four appears is settled before the invitation, because the invitation names it. When you choose, we also hold the date; then the invitation goes out.
- 04
The audio team
How your mixing desk sends the stage microphones to our laptop and takes the voice of the AI avatar back is the longest item, about two weeks. It sets how early you should send your request.
- 05
The briefing
Topic, guests and its position come last, because the lead time is taken up by the technology. The briefing has to be ready by the dress rehearsal at the latest, because that is where the AI avatar already answers on your topic.
02The name
One line of the briefing moves up front
If your invitation names the character, it fixes that name early. The name is also the cue the AI avatar answers to. If someone on stage has a similar name, the character gets a different one, and the invitation should already carry it. So hold your guest list up against its name while the character is being chosen, before the first invitations go out.
Send your request while the invitation is still a draft, even if not every guest has confirmed. Then the paragraph about the AI avatar goes straight in, with the right name.