Creative Studio · Module
Avatar Studio
Disclosed AI presenters: synthetic by default, consented when real
Avatar Studio gives brands disclosed AI presenters: synthetic brand avatars by default and consented representatives of real people within scope. Videos render from scripts in English and Indian languages with provenance, abuse prevention, bulk personalisation with opt-out and costs in the ledger.
What it does
Avatar Studio is for presenter-led video without a camera crew: product explainers, onboarding intros, reminders and personalised sales videos. Avatars are always disclosed as AI.
Synthetic by default (lite, planned W2)
- Synthetic brand avatar library (Class A) with looks, outfits, backgrounds and brand colours, each with an origin record marked synthetic.
- Brand assignment: an avatar belongs to one brand and can be used only in that brand's workspaces or client sub-account.
- Voice assignment with licensed TTS voices per language.
- Script to avatar video: type a script or use one from Content Hub, choose styles and emotions within brand guidelines, and render through approved providers.
- Captions and formats in 9:16, 1:1 and 16:9 through the Video Factory pipeline.
- Disclosure, provenance and usage log: visible AI label, provenance record, C2PA where feasible and a usage audit per avatar.
- Persona policy enforcement: a disclosure rule must exist for the channel and jurisdiction before render and publish.
- Lookalike, public-figure and voice-match checks on uploads and on generated faces and voices.
- Bulk personalised avatar videos with name, company and offer tokens from a CRM segment, checked per recipient for consent basis, opt-out and frequency caps.
- Human review for flagged scripts: political, financial and health claims go to review before render.
- Abuse report and takedown with a report channel, workflow and evidence retention.
- English, Hindi and Tamil, with lip-sync switched on per language only after its eval passes.
Consented real-person avatars (planned W3)
A founder, ambassador or volunteer employee can become a Class B representative avatar only after verified identity and an e-signed, revocable consent that sets scope: tenant, channels, languages, geography, duration and script categories, plus compensation terms and expiry. In Year 1 this is limited to a feature-flagged pilot with design partners after the legal facts are verified and counsel signs off.
Revocation marks the avatar revoked, blocks queued renders, cancels scheduled posts and creates takedown tasks. The consenting person sees every render and use in a usage portal and can revoke there.
Always blocked
Public figures without a licence, minors, deceased persons without estate rights, customers and impersonation are blocked in every wave. Agents can render Class A avatars under caps, but publishing a real person's likeness is an R3 action that only a human can approve.
Features
What Avatar Studio includes
Synthetic brand avatar library
Class A avatars with looks, outfits, backgrounds and brand colours, each with a synthetic origin record.
Brand assignment
Each avatar is owned by one brand and usable only in its workspaces or client sub-account.
Voice assignment
Licensed TTS voices per language; consented voices from W3.
Script to avatar video
Script from Content Hub or typed, with styles and emotions within brand guidelines.
Captions and formats
Captions and 9:16, 1:1 and 16:9 outputs through the Video Factory pipeline.
Disclosure, provenance and usage log
Visible AI label, provenance record, C2PA where feasible and a usage audit per avatar.
Persona policy enforcement
Render and publish are blocked unless a disclosure rule exists for the channel and jurisdiction.
Lookalike, public-figure and voice-match checks
Checks at upload and on generated faces and voices.
Bulk personalised avatar videos
Tokens from a CRM segment with per-recipient consent basis, opt-out exclusion and frequency caps.
Cost estimate, caps and ledger
Estimate per render and bulk job, an avatar minutes meter and rate limits for new tenants.
Human review and R3 publish gate
Sensitive scripts are reviewed before render; publishing a real person's likeness needs human approval.
Abuse report and takedown
Report channel, takedown workflow and evidence retention.
Class B consented avatars and revocation
Verified identity, scoped revocable consent, usage portal for the person and full revocation propagation (W3).
Agent-ready avatar packages
Disclosed avatar assets for the Website Chat Greeter and WhatsApp video replies (W3).
Workflow
Render a synthetic avatar video
- 1
Choose a brand avatar and a licensed voice
- 2
Write or import the script; sensitive scripts go to human review
- 3
Confirm a disclosure rule exists for the channel and jurisdiction
- 4
Review the estimate and render
- 5
Label, provenance and usage log are attached
- 6
Approve and hand off to the owning app; avatar minutes post to the ledger
Workflow
Bulk personalised videos
- 1
Select a CRM segment and map name, company and offer tokens
- 2
Tokens are sanitised and checked for offensive or injected text
- 3
Recipients without a consent basis or who opted out are excluded and frequency caps applied
- 4
Confirm the bulk estimate and render
- 5
Approved videos are handed off to Marketing or WhatsApp for sending
AI
AI skills and Digital Employees
Script to Class A avatar video
Renders synthetic presenter videos with licensed voices.
Lip-sync per language
Applies lip-sync only for languages whose eval has passed.
Lookalike and public-figure checks
Screens faces and voices before every render.
Sensitive-script classifier
Routes political, financial and health scripts to human review.
Consent-scope compliance check
Checks a Class B script's category, channel, language and geography against the grant (W3).
Digital Employees that use this module
- Creative Variant Generator
- AI Video Producer
- AI SDR — Inbound Lead Responder
- Website Chat Greeter
- Client Reporting Agent
They act through the same governed Actions as people; risky steps wait for approval.
Frequently asked questions
Are avatar videos labelled as AI?
Always. A visible AI label and a provenance record are attached, and a render cannot run unless a disclosure rule exists for the channel and jurisdiction. Disclosure cannot be switched off.
Can I make an avatar of our founder?
Consented representative avatars are planned for W3 at scale. They need verified identity and a scoped, revocable consent. Year 1 allows only a gated design-partner pilot after legal facts are verified and counsel signs off.
What happens if someone withdraws consent?
The avatar is marked revoked, queued renders are blocked, scheduled posts are cancelled and takedown tasks are created for published videos, all with an audit trail.
Can I send personalised avatar videos to a customer list?
Yes, from W2 in lite form. Each recipient is checked for consent basis, opt-outs are excluded and frequency caps apply. Sending is done by Marketing or WhatsApp, not by Avatar Studio.
See Avatar Studio in your workflow
Tell us about your team and the outcomes you want. We will walk you through the apps, the Digital Employees and the plan for your industry.