Create an editable assisted reply selection promo video for [PROJECT] and [AUDIENCE]. Build a new result from [APPROVED_CONTENT] and [BRAND_RULES], using the credited reference to discuss composition and timing.

## Required Content and Inputs

Introduce an approved conversation context; run a labeled sample reply-generation step; compare a small set of candidate replies; show the human selecting and editing one.

Ask for exact on-screen copy, authorized assets, intended destination, output dimensions and licensed audio. Identify missing task data before implementation. Keep real claims and sample data distinct; do not invent customer results or capabilities.

## Scene Plan

1. Introduce an approved conversation context. Define the visible action, its result and the transition anchor.
2. run a labeled sample reply-generation step. Define the visible action, its result and the transition anchor.
3. compare a small set of candidate replies. Define the visible action, its result and the transition anchor.
4. show the human selecting and editing one. Define the visible action, its result and the transition anchor.

Return a timing table with scene start and end, copy, assets, motion and audio cues. The reference is 20.1 seconds at 1920 × 1080; adapt the timeline to the supplied content instead of forcing unreadable text into that duration.

## Motion and Visual System

Keep source context and generated candidates in separate records. Show the review step explicitly rather than animating generated text directly into a public posting state.

Use the selected project design system. Establish final readable states before animating entrances. Preserve image and logo proportions, separate editable type from raster art and use a restrained hierarchy for supporting labels. Keep one recognizable anchor between adjacent scenes.

## Implementation and Audio

Use the existing project stack and an explicit timeline that can be inspected at any timestamp. Store copy, data, colors, scene timing and motion parameters in editable configuration. Seed any randomness so the same timestamp renders consistently. Do not depend on wall-clock timers to define an export frame.

Map supplied music or sound cues to timeline markers. If audio is missing, deliver a silent timing preview first. Do not claim beat or voice synchronization without the actual authorized track.

## Review Priorities

- Check that quoted context is authorized and that the chosen reply answers it. Do not claim a specific number of successful replies or automatic posting unless those facts are supplied and supported.

Inspect the longest real label, the first and last frame and both sides of each scene boundary. Check phone-size reading, foreground occlusion and the actual output aspect ratio. Compare the exported duration, frame count and audio tail with the approved timeline.

## Deliverables

Return editable source, the timing table, one still per scene, a poster and an MP4. Provide the low-cost scene preview before the full render. Distinguish a pre-rendered interface demonstration from working application behavior.

Reference creator: @robj3d3.
Source: https://x.com/robj3d3/status/2103949734520279060