Researchers introduce PyRUA-DataFarm, a system that replaces human teleoperators with a coding LLM agent to generate real-ready robot training data at lower cost. Given a single calibration episode and a data request, the agent reconstructs the task in simulation by writing the scene description, an expert controller, and a success judge; independent judges then certify the expert and each rendered episode. A video model conditioned on real images renders synchronized, multi-view photographic video for every simulated rollout, labeling frames with exact robot actions so policies can be trained directly and deployed on hardware. On five real-robot tasks, policies trained exclusively on farm data achieved 93 successes in 100 trials versus 84 for policies trained on an equal number of teleoperated episodes. Under distribution shift (unseen tablecloth and colored lighting), farm-trained policies maintained 87 successes while teleoperation-trained policies dropped to 46. Adapting a new task is fully automatic, costing approximately $21 in API usage (one-time per task) plus $5 for a render contract; each additional episode averages $0.80 for simulation, rendering, and review with no robot or operator time required. One hour of three-view data costs roughly $107 on average, slightly below a $117 teleoperation quote in China and well below $290 in North American labor costs alone.