A DIRECT-driven, active-learning workflow that autonomously builds machine-learning interatomic potentials (MLIPs) for batteries, ferroelectrics, catalysts, polymers, and electrolytes, fully integrated with the atomate2/jobflow ecosystem.