How to create a Custom AI Team Member – Claymind

A Custom AI Team Member lets an approved person from your business appear as an on-screen presenter in Bizrie videos. The finished AI Team Member learns its appearance, expressions, posture, and mannerisms from the training footage, so the quality of the recording directly affects the result.

Use this guide before recording or submitting any footage.

What you need to submit

AI Team Member name: The name you want displayed in your Bizrie workspace.
One continuous training video: MP4, WebM, or MOV; between 1 minute 45 seconds and 2 minutes 30 seconds; up to 250 MB.
One separate consent video: MP4, WebM, or MOV; between 3 and 60 seconds; up to 75 MB.
Permission confirmations: You must confirm that the person shown explicitly authorizes creation and use of their likeness, understands that Bizrie and its contracted video providers will process the footage, and knows how to withdraw permission.

The same real person must appear in both videos. Do not submit footage of anyone who has not knowingly agreed to become a Custom AI Team Member.

Set up the camera and room

Camera: A recent phone, webcam, or camera can work. Record in 1080p when possible and use a tripod or stable mount.
Framing: Use landscape orientation. Frame the person from approximately the chest up, centered, at eye level, with the entire face visible.
Background: Choose a clean, still background. Avoid people walking behind the subject, moving screens, reflections, and changing light. If possible, stand several feet away from the background to reduce shadows.
Lighting: Use soft, even light from the front or both sides. Avoid harsh overhead shadows, bright windows behind the person, flickering lights, and strong glare on glasses.
Sound: Record in a quiet room without music, fans, echo, other speakers, or notification sounds. The finished clips use an AI-generated voice, but clear training audio is still important for natural timing and review.
Clothing: Wear solid or simple colors. Avoid large logos, text, busy patterns, highly reflective jewelry, and anything that covers part of the face.

Make a short test recording first. Confirm that the face is sharp, skin tone looks natural, exposure does not change, and the person remains inside the frame while making a small gesture.

Record one continuous performance

Do not cut, splice, speed up, slow down, or combine multiple takes. Record the following three parts in one continuous shot:

1. Listening — about 15 seconds, silent

Look directly into the camera as if listening to another person.
Keep the mouth closed at rest.
Use a subtle smile, one or two small nods, and an occasional gentle eyebrow movement.
Stay engaged without speaking or making dramatic expressions.

2. Speaking — about 90 seconds

Speak clearly and naturally to the camera at a steady pace.
Keep consistent eye contact with the lens.
Pause for one or two seconds between longer sentences with the lips closed.
Use a warm, confident expression and natural variation in delivery.
If using gestures, make approximately four or five slow, subtle gestures below shoulder height and keep both hands inside the frame.

You can speak about your business, role, services, or another familiar topic. The exact words will not become the final script; this section teaches the AI Team Member how the person naturally speaks and moves.

3. Idling — about 15 seconds, silent

Finish in a calm, attentive pose while looking into the camera.
Keep the mouth closed with a neutral expression or subtle smile.
Breathe normally and use only an occasional small nod.
Do not end with a sudden turn, wave, or movement out of frame.

Body language and expression guidance

Posture: Sit or stand upright with relaxed shoulders and an open, natural stance.
Head movement: Remain mostly front-facing. Keep turns small, slow, and controlled.
Hands: Gestures are optional. When visible, keep them below the shoulders, away from the face, and within the camera frame.
Facial expressions: Use natural smiles, attentive listening, and mild expression changes. The recording should feel friendly and professional rather than exaggerated.
Eye contact: Look into the camera lens throughout the recording instead of watching yourself on the screen.

Avoid: pacing, swaying, fidgeting, quick head turns, repeated large gestures, crossing in and out of frame, touching or covering the face, using props, coughing, looking at notes off-camera, open-mouth resting poses, or extreme facial expressions. Unusual movements or expressions in the source footage may be reproduced by the AI Team Member.

Record the consent video

The consent video must be a new, real recording of the same person shown in the training footage. Their face and voice must be clear. Do not upload a screen recording, an AI-generated person, or a recording of someone else reading consent.

The person can say:

“I, [full name], consent to Bizrie and its contracted video providers using these recordings to create and operate a Custom AI Team Member based on my likeness. I understand that I can withdraw my permission and request deletion of my source footage.”

Speak slowly enough for every word to be understood. Record in clear lighting and keep the video between 3 and 60 seconds.

Submit the files in Bizrie

Open AI Team Members in your Bizrie workspace.
Select Custom, then choose Create custom.
Enter the AI Team Member name.
Upload the continuous training video and the separate consent video.
Review and select the permission, processing, and withdrawal acknowledgments.
Select Submit custom AI Team Member.

What happens after submission

Custom creation is not instant. Bizrie reviews the footage, submits approved footage for provider processing, provisions the AI Team Member, and performs a quality check. Your workspace displays the submission status and makes the AI Team Member available after it is approved and provisioned.

If the recording is blurry, dark, edited, unstable, missing a clear face, uses a different person for consent, or contains unclear audio, you may be asked to record it again. The Custom add-on currently includes one training or retraining submission per billing period, so review the checklist carefully before submitting.

Final checklist

The same consenting adult appears in both recordings.
The training video is one continuous 1:45–2:30 shot with no edits.
The sequence includes listening, speaking, and idling.
The face remains visible, centered, sharp, and evenly lit.
Eye contact, posture, gestures, and expressions are natural and controlled.
The background is still and the audio is clear.
The consent statement is understandable and the consent video is 3–60 seconds.
The files are MP4, WebM, or MOV and within the upload size limits.

To withdraw likeness permission or request deletion of stored source footage, use the Custom AI Team Member page in Bizrie or contact support@bizrie.com. Existing generated clips must be removed separately.