Volunteer recruitment video from an AI avatar: 25-second clip

A volunteer ask is a good fit for a scripted avatar clip because it is one request repeated to many people: what the job is, how long it takes, and where to sign up. A 25-second clip is enough. It will not replace the coordinator who answers questions, so end by pointing to a person or a form.

One scene or three

A single script gives you one continuous take. If you want a hook, a short pause and a closing ask, use video_inputs instead, with ordered scenes. A scene can be spoken (voice.type: "text") or a silence beat (voice.type: "silence" with a required duration). The current execution supports one avatar for each final video and expects scene backgrounds to resolve to one shared scene, so the three scenes are the same person in the same place, not a cut between locations.

Script skeleton

• Hook (about 3 s): the specific job, such as Saturday pantry sorting.
• Detail (about 15 s): the time commitment and who it suits, in plain terms.
• Ask (about 5 s): where to sign up, and the name of a real contact person.

Captions

Many viewers watch muted, so turn on inline captions: "captions": { "enabled": true, "style": "slam", "language": "auto" }. The captions are burned in after generation using the spoken text, and a failure in that stage is soft: the clip can still succeed with captions.status=failed. Inline captions are limited to videos whose estimated duration is 60 seconds or less, which a 25-second ask clears easily.

Cost for a recruiting season

Prices are per second at the fixed rate card. A season of four clips (one per role) at 25 seconds each:

A note on trust

People give time to people. Say in the page text that the video is AI-generated and that a named coordinator will reply. If a real member appears as the avatar's face, get their consent first.

Reusing one avatar across roles

Make one presenter and use the same handle for every role clip. The handle is stored without a leading at sign, and you pass it as avatar_handle in each talking-video request. Changing only the script and the scene prompt gives you a set of clips that look like one organisation's channel. Keep a spreadsheet of handle, role and script version so that a coordinator can request a re-render when shift times change; the avatar does not need to be recreated and costs nothing more to reuse than the per-second render.

If a role is sensitive, such as working with children, put the screening steps in the page text and not in the clip, and have a named person confirm the wording.

Before you post it

Check the first-frame preview if you are using a scene photo of your own hall or pantry, since a cluttered photo makes an odd backdrop. Then read the transcript aloud against the sign-up page: dates, times and the contact name must match. Post the clip with a text summary, and note the date the roles were last confirmed so that volunteers are not sent to a filled shift.