Let's see how the MAP function works.

Modern vision-language models allow documents to be transformed into structured, computable representations rather than lossy text blobs.