AI Media
Avatar / talking head
HeyGen renders the performance. EditForge owns the brief, the cut linkage, and the rubric gate.
Stage
Avatar output
Completed avatar renders appear here. Until a job returns media, the stage stays honest — no placeholder talking head implied as live.
Design source · signal
Honesty mark: mock paths never invent face or performance media. Simulated runs are labeled. Live requires provider credentials and still respects the rubric before master.
Projects2
TSWS cold open — avatar bed
draft“Where are we today?”
TQO presenter teach via HeyGen
ready“Three steps. No hype.”
Provider flow
- 1POST /v3/videos {type, avatar_id, voice_id, script} → video id
- 2GET /v3/videos/{id} until completed or failed
- 3Completed → video_url recorded on the job as its result
- 4Attach the MP4 to an EditForge cut
- 5Rubric pass before master ship — the gate is not optional
Brief
Submitting the same brief twice returns the same job — the key avatar-ec9bdd7b is derived from the brief, not the click.