Opus 5.5 is good at explainer videos
Points and comments are a snapshot, not live.
LaunchVideo uses Claude Opus 5.5 to generate explainer videos from a URL or prompt.
LaunchVideo turns a URL or prompt into an MP4 explainer video. A serverless agent on OpenComputer uses Anthropic's Claude Opus 5.5, which writes the film as HTML and JavaScript. The agent has three tools: web_fetch (extracts page text, colors, fonts), check_scene (loads film, reports JS errors), and render_video (renders via headless Chromium to ffmpeg at 1920x1080 30fps). A virtual clock makes rendering deterministic. Each video takes about four minutes and roughly 100k tokens. The agent deploys as a single TypeScript file with no external queue or server. Example outputs: NVIDIA, Linear, Infera.
What commenters are saying
Several commenters dismissed the output as 'slop' comparable to pre-AI explainer videos, questioning whether anything of value was lost. A parallel thread debated whether LLM-wrapper SaaS businesses are viable, with some arguing they are easily replicable while others noted market incumbents often ignore such threats. One commenter described how ChatGPT replaced their niche cocktail app entirely, illustrating the vulnerability of narrow tools to general-purpose AI.