AI presenter for a SaaS demo video, with or without a product image

A SaaS demo presenter on Sume is one POST /v1/avatar-1.0/talking-video call with a ready avatar and a script. A SaaS product has no physical item to hold, so omit product_image and you get a productless avatar video. Each job must estimate at 4 to 60 seconds, so a longer demo is several jobs.

HeyGen's updates feed lists BrandKit and Video Podcast among its recent announcements; this post covers only the Sume route. Facts are from the avatar video docs, read 2026-10-01.

When do I add a product image?

product_image is optional. Leave it out for a presenter talking over a scene, and add it when a physical or packaged item should appear in the shot. A screenshot of your UI is a different job: use a photo scene (scene: { "type": "photo", "image_url": "https://..." }) as a scene reference, or a prompt scene for direction. Media URLs must be fetchable public HTTPS.

How long can the demo be?

Scripts and multi-scene plans are accepted when the estimated duration is 4 to 60 seconds inclusive. Shorten the script or split it into several jobs. With ordered video_inputs you can compose hook, demo and silence beats in one video, but the total must still land in the same window. If you turn on inline captions, an estimate above 60 seconds is rejected.

What does it cost?

Avatar video is billed per second and the rate depends on quality (standard, plus default, max) and whether a product is present.

What should I do for a five-minute demo?

Split the script into scenes of under 60 seconds, make one job per scene, and join them with Timeline 1.0; the steps are in avatar video longer than 60 seconds.