New product launch video with an AI avatar presenter

A new product launch video is a short announcement that names the problem, shows what is new, and ends with one call to action. Keep the main cut under a minute for social feeds, and make a wide version for the launch page.

A presenter-led version can be made from the approved launch copy on release day: an AI avatar reads the script with your product image in frame, rendered once per shape you need. Facts come from Generate avatar video and Avatar video previews, read on 2026-09-28, plus Sume's current code where noted.

What goes in a product launch video script?

Write it from the same approved copy as the launch page, so the video says what the page says:

• Hook: the problem, in one line.
• What's new: the product and the one change that matters most.
• Proof: one concrete detail from your release notes.
• Call to action: one action, and where to take it.

How do I make a launch video with an AI avatar?

Send the script with the avatar's handle, a product_image URL, and a scene prompt for the setting; media fields must be fetchable public HTTPS URLs. AI avatar video with your product and background covers both fields, including a photo scene. For a hook, a silent beat, and a call to action as separate scenes, use multi-scene video_inputs instead of script; the multi-scene avatar video API shows the scene fields.

curl -X POST https://api.sume.com/v1/avatar-1.0/talking-video \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: launch-trail-bottle-9x16-v1" \
  -d '{
    "avatar_handle": "launch_host",
    "script": "Tired of bottles that leak in your bag? Meet the new Trail bottle. Order it today.",
    "product_image": "https://example.com/launch/product.png",
    "scene": { "type": "prompt", "prompt": "Clean studio with soft daylight" },
    "aspect_ratio": "9:16",
    "quality": "plus"
  }'

Will the product look right in the video?

The docs call product_image an optional product/reference image, so treat it as a reference and check how it lands before launch day. Start from an avatar video preview: it returns first-frame stills without starting the full render, and when they look right you start the video from the preview id.

How do I make a version for each channel?

Render once per shape. aspect_ratio supports 1:1, 3:4, 9:16, 4:3 and 16:9 (default 9:16). Each video is its own job and bills on its own. aspect_ratio is a structural field for previews too: changing it requires a new preview, so preview each shape you plan to publish.

What are the limits, and what does it cost?

• Each video covers an estimated 4-60 seconds; a longer launch film needs several clips joined.
• resolution is currently 720p, with one presenter per video.
• In current code the avatar route speaks English only.
• No screen recordings: for a software launch, the product walkthrough needs another tool.