Puvexa_AI

OpenAI Dots can now run a motion design team. three of them made this film in After Effects from one message how the job ran: brief → one message starts everything. "team, build a launch film for plotline in After Effects. from scratch." no storyboard, no shot list split → every Dot claims its part in the chat before it opens anything. one takes the opening and research scenes, one takes animation curves and camera moves, the third takes visual styles and 3D titles build → all three work in the same project at the same time, each from its own cloud computer. three cursors move through the timeline, the comp viewer and the effects panel in parallel footage → Higgsfield generates the shots and the episodes, and the Dots cut them into scenes without any stock clips or manual imports check-ins → progress shows up in the chat with real numbers. "prompt scene is up · 84 layers", "research scene done · 137 layers", "3D titles with an orbit camera · 206 layers". nobody has to open the project file to follow along a preview renders a few seconds behind the build, so the film plays while it is still being made. before export, the curves Dot runs a camera pass on every scene, then the master gets assembled with music, rendered in ProRes and dropped into the chat as a file three agents split one brief like a real motion team and ended up with 206 layers and around 2,000 keyframes how Dots compares to Grok Bot and a Sonnet 5.5 studio on the same kind of job is in the article below ↓