FRST AI Learning Track

FRST AI Learning Track is a full-scale educational content production pipeline built entirely with generative AI — designed to replace an ≈R$80K traditional production workflow without sacrificing quality, consistency, or human creative control. My role covered the entire production stack: creative direction, visual direction, AI systems design, and quality control. The pipeline works like this: the subject-matter expert writes the script. From there, an automation I built takes over — running the script through a visual direction layer that defines shot type, framing, and scene style (medium shot, three-quarter, wide, realistic, or 3D avatar) for every single scene. That output feeds into a full visual breakdown, generated automatically using my own audiovisual knowledge base as the foundation. The scenes are then rendered inside a platform built in partnership with the FRST team, using a realistic 3D avatar I created from the expert's likeness — paired with a cloned voice built from his own recordings. Every generated scene goes through human curation: I review each one, validate or reject, and regenerate when needed. After curation, I assemble the full edit, add the cloned voice, and take the cut into After Effects for intros, lower thirds, motion graphics, and finishing — until the lesson is complete and ready to publish. What started as a single module is now being scaled into full learning tracks produced at speed, without a camera, without a studio, and without a crew.