AI Agent Makes YouTube Videos for Under $2: Zero Human Input

An AI agent is creating vertical videos for YouTube at a cost of less than two dollars for five videos, with zero human involvement. This automated system, housed in a datacenter, generates video ideas, writes scripts, creates German voiceovers and music, and renders 30-second videos. The process involves seven steps, with only three using a large language model, GPT-5.6 Terra Pro. Other steps include voice synthesis and music generation, along with deterministic code for rendering and uploading. A key design philosophy is "judgment before spend," meaning creative steps, which are inexpensive, happen before costly production steps. This prevents spending money on off-brand videos. The creator initially tried having the LLM generate video visuals, which proved expensive and unreliable due to issues like contrast problems and API failures. This system demonstrates how building automated systems can lead to efficient content creation.