A virtual brand ambassador for a B2B company is one synthetic presenter, a person who does not exist, who appears in your product explainers, launch clips and event openers, always with the same face, the same wardrobe rules and the same voice. On an 8frame board that is an identity sheet made with image models, reference-to-video models for the presenter in motion, lip-sync models for talking shots, and a register of where each clip runs and how it is labelled. The limit people do not expect: when the ambassador appears in an advertisement, New York already requires a conspicuous disclosure that a synthetic performer is in it, and California signed a similar rule in September 2026.
This page is the B2B version. For the general idea of a brand character, see what a brand board is and our page on making a brand mascot with AI.
TL;DR
| Step | Node or model on 8frame | Fixed price in credits | Input it needs |
|---|---|---|---|
| Candidate faces | Nano Banana 2 / Seedream 5 Lite | 8 to 22 by size / 6 | a written spec |
| Identity sheet | Nano Banana Pro or Seedream 5 Lite | 21 at 1K or 2K / 6 | the chosen face as a reference |
| Presenter in motion | Happy Horse reference-to-video | 95 for 5 s at 720p, 189 at 1080p | 1 to 9 character references |
| Presenter in motion, alternative | Grok Imagine reference | 35 or 48 for 5 s, by resolution | 1 to 10 reference images; audio always on |
| Talking shot | Pruna Avatar | 720p 3.375 per second, 1080p 6.075 per second | portrait plus audio, up to 35 s |
| Voice | ElevenLabs TTS | 13 per started 1,000 characters | script text |
| Assembly | Video Editor timeline | none | clips, voice, captions, label |
We quote price and settings for these models, not results.
The board, node by node
There is no published template for this yet; the CHARACTER POSES template on /workflows/character-poses opens as a board you can duplicate for the pose step. The rest is below in words.
1. Write the ambassador as a brief
A B2B ambassador is a role, not a celebrity. Write it on a sticky note: what they explain (product, onboarding, events), the tone (calm, direct, a little dry), age range, style of dress that fits your buyers, and what they never do (no jokes about competitors, no medical or financial advice, no claims the product team has not approved). Give them a first name only, and make sure it is not the name of a real person in your field.
2. Generate candidates, then lock one face
Generate a spread of candidates from the brief on an image node: Nano Banana 2 (8 to 22 credits by size) or Seedream 5 Lite (6). Check the shortlist for resemblance to real people, especially well-known people in your market: an ambassador who looks like a competitor's founder is a problem you made yourself. Pick one face.
3. Build the identity sheet
From the chosen face, make the sheet the whole team will work from: front, three-quarter left and right, profile, full-length standing, seated, two outfits from your wardrobe rules, two expressions. Seedream 5 Lite takes up to 14 reference images and Nano Banana 2 up to 10, so the face and the outfit references can go into every node. Keep the sheet in one frame on the board with a "do not edit" sticky. Every later node takes its references from this frame, never from an earlier experiment.
4. Presenter in motion
For shots where the ambassador walks, gestures or turns to a screen:
- Happy Horse, reference-to-video with 1 to 9 character references, 95 for 5 s at 720p and 189 at 1080p. Attach the identity sheet images.
- Grok Imagine reference mode with 1 to 10 images, 35 or 48 for 5 s by resolution. Audio is always on, so the clip comes with generated sound; mute it on the timeline.
We have not measured which holds identity best. Make the same shot on both from the same references and keep the take that matches the sheet.
5. Talking shots from one voice
Pick one ElevenLabs voice for the ambassador and write it on the board (13 credits per started 1,000 characters, up to 5,000 per generation). For on-camera speech, wire a portrait from the identity sheet and the line's audio into Pruna Avatar (up to 35 s), OmniHuman v2 (71 for 5 s, billed per second of audio, 4 to 35 s) or Hailuo H3 Max Lip Sync (about 14.8 s of audio, 34 to 216 per 5 s by resolution). 8frame has no voice cloning and no tool that re-syncs the mouth on existing footage, so make talking shots from the audio, not the other way round.
6. Assemble, caption, label
On the Video Editor timeline, put the clips, the voice layer and captions ("Generate captions" from the pasted script, timed at an estimated 160 words per minute). Add the disclosure as a text layer where it will be seen, not only in a description. Do not use Montage for talking clips: its merged file has no sound.
7. Keep a register
A frame on the board, or a sheet next to it: every published clip, where it runs, whether it is an ad, the date, and the disclosure used. When someone asks in a year whether the ambassador was labelled in the trade-show ad, the answer is on the board.
The disclosure rules for ads
Two US states have written rules for synthetic people in advertising. General information, not legal advice:
- New York, General Business Law section 396-b. A person who produces or creates an advertisement for a commercial purpose, with actual knowledge that a synthetic performer is in it, "shall conspicuously disclose in such advertisement that a synthetic performer is in such advertisement." A synthetic performer is "a digitally created asset created, reproduced, or modified by computer" meant to give the impression of a human performance and "not recognizable as any identifiable natural performer." Civil penalties are "one thousand dollars for a first violation" and "five thousand dollars for any subsequent violation." The section does not apply to audio ads, to AI used solely to translate a human performer, or to promotion of expressive works such as films and games when the use matches the work.
- California, SB 1050, signed on 16 September 2026 and applying from 1 January 2027. The Governor's announcement says it requires "explicit disclosure on any video or audio advertisement" that uses AI-generated performers, defines a synthetic performer as "An AI-created digital figure, voice, or representation", and "prohibits the continued use of any advertisement found to be in violation of the law."
Two practical consequences for a B2B ambassador. First, California's definition includes a voice, so an audio ad with the ambassador's TTS voice is in scope there, while New York's section excludes audio ads. Second, both rules are about advertisements; an internal training video is not an ad, but a LinkedIn promotion of the same clip is. The safe habit is to label every public clip.
Which model for which step
| Need | Model | Limit |
|---|---|---|
| Many references in one image node | Seedream 5 Lite | up to 14 references, 6 per image |
| Hero portrait for the sheet | Nano Banana Pro | 21 at 1K or 2K, 42 at 4K |
| Ambassador moving, many character refs | Happy Horse | 1 to 9 refs; 95 at 720p, 189 at 1080p for 5 s |
| Ambassador moving, lower price | Grok Imagine reference | 1 to 10 images; audio always on |
| Talking up to 35 s | Pruna Avatar, OmniHuman v2 | priced per second of audio |
Where it breaks
- Identity drift. Over dozens of clips, the face moves. Compare each clip with the sheet, not with the last clip.
- Resemblance. A synthetic face can resemble a real person by accident. Check before you lock it, and change it if someone recognises a likeness.
- Generated sound. Grok Imagine always adds audio; Happy Horse files came with an audio track in our measurement. Mute and use the ambassador's TTS voice.
- Claims. The ambassador says only what your product and legal teams approved. A synthetic presenter does not make a claim less yours.
- Labels. A label in the caption that gets cut on reposting is not conspicuous. Put it in the video.
Who owns the ambassador internally
A synthetic presenter needs an owner the way a brand guideline does. Name one person who approves every new clip, keeps the identity sheet and the voice card, and maintains the register. Without one, three teams make three slightly different ambassadors within a quarter. Write the owner's role on the board next to the sheet, along with the rules for the ambassador: topics they cover, claims they never make, the disclosure wording, and where the label goes in the frame.
FAQ
What is a virtual brand ambassador? A synthetic presenter, a person who does not exist, used across a brand's videos with one fixed face, wardrobe and voice. On 8frame it is built from an identity sheet and reference-to-video and lip-sync models.
Do I have to label an AI brand ambassador in ads? In New York, ads with a synthetic performer need a conspicuous disclosure under General Business Law section 396-b. California's SB 1050, signed on 16 September 2026 and applying from 1 January 2027, requires disclosure on video and audio ads with AI-generated performers. Label every public clip.
Which model keeps the ambassador's face across clips? Happy Horse takes 1 to 9 character references and Grok Imagine 1 to 10 reference images. We have not measured which holds identity better; test both on the same references.
Can the ambassador speak? Yes, through a TTS voice and a lip-sync model from a portrait: Pruna Avatar, OmniHuman v2 or Hailuo H3 Max Lip Sync. There is no voice cloning and no mouth re-sync on existing footage.
Can I base the ambassador on a real employee? Then it is not a synthetic performer but that person's likeness, and you need their written agreement for every use. Many teams keep the two apart: a real spokesperson films, a synthetic ambassador is fully generated.
Sources
- New York State Senate, General Business Law section 396-b (disclosure duty, definition of synthetic performer, $1,000 and $5,000 penalties, exemptions), checked 2026-11-09
- Office of the Governor of California, Governor Newsom signs new law to protect workers, require disclosures on AI-generated advertising (SB 1050, 16 September 2026), checked 2026-11-09
- 8frame prices, defaults and limits: the 8frame app, checked 2026-11-09
Try it without signing up: the 8frame canvas opens on a real board. The canvas is free and unlimited; generation is paid from $19/month.