Checklist for an AI avatar at your event, from the first request to the photo caption

01Before the invitation

Who waits on whom before the invitation goes out

The invitation waits on two small decisions. It can only say what the AI avatar listens to once you know whether a room microphone will be added. And if your privacy notice names a deletion period, you need to have agreed on it with us first.

• Organizer: get in touch with a date and a format, and put us in contact with your sound engineer.
• Organizer: choose the character and check its name against the names of the guests on stage, since that name is what it listens for; if someone there has a similar name, the character gets a different one.
• Organizer: decide whether the screen labels it as an AI avatar for the whole appearance.
• Organizer: decide whether a room microphone is added. By default it hears only the stage microphones. Voices from the audience reach the transcript only if a room microphone is on the aux send.
• Organizer and Aura51: agree on the period after which the transcript, the briefing and the test recording are deleted. The transcript sits on our laptop.
• Organizer: mention in the invitation to guests and moderator that an AI avatar will join the discussion and which microphones it listens to. Waiting for the briefing would be too late for that.
• Sound engineer: confirm a dedicated microphone for the moderator on its own channel, the aux send with the other stage microphones, return path, screen and a seat for us at the desk, plus the room microphone if you want one.

02Briefing

The dress rehearsal waits on the briefing

At the dress rehearsal, the character already answers on your topic. By then it needs to know who sits on the panel with it and which position it takes. It does not need any of that sooner.

• Organizer: send the briefing, with the topic, the participants and their roles, and the character’s position.
• Aura51: enter the names from the briefing on the tablet; it uses them to address the guests correctly.

03Rehearsal

At the dress rehearsal, everyone practices their part once

This is where the sound engineer, the moderator and one of us sit together. Preparation also includes a short test recording, so it recognizes its name even in an unfamiliar pronunciation.

• Sound engineer: check that the moderator’s microphone arrives on its own channel, that the aux send carries the other stage microphones but not the avatar’s voice, and that its voice reaches the room through the return path.
• Sound engineer and Aura51: listen for whether the avatar’s voice finds its way from the room speakers back into the stage microphones.
• Guests on the panel: say a few sentences into the microphone before the event. Only then do we match each voice; once, when we matched too early, the speech recognition took two voices for one.
• Moderator: call on it by name, then follow up without the name right after its answer.
• Moderator: tap STOP once. At the event, the moderator stops it with this button.
• Moderator: say the line for an internet outage out loud once.

04On the day

On the day, we set up what worked in rehearsal

• Aura51: connect the laptop and audio interface next to the desk, switch on the hotspot and hand the tablet to the moderator.
• Sound engineer: set the desk as in the rehearsal, then run one test call before the room fills up.

During the appearance, the moderator does nothing that was not already done once in rehearsal: call on it, follow up, stop it, say the line for an outage.

05Afterwards

After the event, it stays an AI avatar in pictures

• Organizer: in photo captions and videos of the event, name the character as an AI avatar, for example “Albert (AI avatar)”.
• Aura51: delete the transcript, the briefing and the test recording after the agreed period.

The first item on this list is your request, with a date and a format. Tell us right away who handles sound for you. That is the person we have the longest conversation with on the whole list.