TAS-AI

TAS-AI is organized as a staged autonomous workflow with agnostic discovery, physics-informed inference, motion-aware sequencing, and an optional audit layer. TAS-AI treats autonomous triple-axis spectroscopy as separate tasks for signal discovery, Hamiltonian selection, and parameter refinement. The workflow begins with agnostic discovery before switching to physics-informed planning after signal structure is localized. The automatic handoff is triggered after discovery identifies a localized signal region sufficient to instantiate a physics model on a restricted action set. The integrated hybrid run used coarse grid sampling and enhanced Log-GP before physics refinement began at measurement 29.