SaaS pricing needs intelligent pricings (iPricings)–machine-readable representations that enable automated validation and analysis. Yet today, pricing is published in heterogeneous, ad-hoc HTML with no widely accepted notation or standard, leaving crucial associations (e.g., plan \(\leftrightarrow \) add-on \(\leftrightarrow \) subscription) tied to positional or visual cues and thus ambiguous for machines. We introduce A-MINT (Automated Modeling of iPricings from Natural Text), an end-to-end LLM-powered engine that converts free-form pricing pages into valid iPricings. A-MINT combines HTML preprocessing, LLM-driven extraction, automated constraint validation, and an LLM-based correction loop to resolve structural and linkage errors. A-MINT achieves a mean F \(_1\) -score of 70.51% on SPECTRA, a dataset containing 162 real-world iPricings from 30 SaaS providers over six years, showing that iPricings can be modeled at scale despite the absence of a standard representation for pricings on the web.