Models are finally useful on messy PDFs

A stack of supplier PDFs is a common reason someone emails us. The files are not the same. Columns drift. Totals hide in footnotes. The person who used to retype them is tired.

What changed this year is not a slogan. The models we test now keep the figure that was printed, even when the layout is ugly. That is the whole job.

We still check the output. The point is that the checking is now the work — not the retyping.