Agentic Video Production *Shots. Editing. Sound. A connected production workflow.* AI video models can now produce a striking clip from a single prompt. Turning clips into a scene is harder: the shots have to match, the cut has to hold, the sound has to fit the picture, and someone has to answer for each choice. Agents are starting to drive, and the options keep multiplying. This talk tours them in practice: * Claude and Astra calling video models through MCP and gateways such as Token360 * the built-in agents in Higgsfield, Magnific, HeyGen, etc... * Claude Motion for motion graphics and animation that stays editable * Claude blocking scenes in Blender and working inside Unreal Engine * Judging a production and driving remediation cycles * a finish in DaVinci Resolve It also covers what it takes to make the process repeatable: story bibles, shared references across episodes, clear handoffs, skill development, many retries. Examples come from a vertical series on the Revolutionary War and from open-source multi-agent systems such as FilmAgent and OpenMontage. You’ll leave with a grounded sense of what agents can and can’t do for video today, and a small test to try on your own work.
who's going
27 going on meetup