Personalized Broadcast Video — how-to

<p>Train a Brand Character LoRA once (or use a saved one), pick a voice (preset ElevenLabs or your cloned voice), drop the merge token into your broadcast subject or body. Pre-render walks every recipient before send, deduping by (script, voice, model) so 1000 recipients with 200 unique first names cost the work of 200 renders, not 1000. Recipients see a ~2-second animated GIF preview directly in Gmail / Outlook before they click — proof that your brand character knows their name.</p> <h2>What it does - Per-recipient talking-head MP4 with first-name-personalized opener - Animated GIF inbox preview (renders inline in Gmail + Outlook) - (script × voice × model) cache dedupe — pricing scales with distinct names, not recipients - Pre-render fan-out + 30-min hard cap so the broadcast never blocks - Cloned voice (ElevenLabs Instant) or curated preset - Same pipeline powers 2-3sec talking-head openers spliced into creatives</h2> <h2>How to use it 1. Start a new Email Broadcast. 2. Add a Brand Character and a script with a {first_name} token. 3. OCCS pre-renders a unique talking-head clip per recipient with an animated-GIF inbox preview. 4. Send — each recipient sees a personalized video.</h2> <h2>Good to know - <strong>Available on:</strong> Starter, Growth, Scale - <strong>Connected tools:</strong> Replicate (flux LoRA + lucataco/latentsync), ElevenLabs (TTS + voice clone) - <strong>Find it in the app:</strong> open `/email/broadcasts/new` - Cost scales with distinct names, not total recipients (identical names dedupe). - Pre-render runs with a 30-minute cap so the broadcast never blocks.</h2>