Build log, April 9, 2025
Build log: table extraction when the table spans four pages
The problem
This one comes up in almost every matter review, and the answer is less settled than the confidence around it suggests.
In testing, the gap between a fast pass and a careful pass was not accuracy on the central question. It was whether the reviewer noticed the interaction with an adjacent provision.
What we do about it
We treat anything the system cannot ground in source text as unverified and show it as such. A visible gap is cheaper than a confident error.
In the current build this runs as part of the standard pass, so it applies to every document in the set rather than only the ones someone thought to check.
Where it breaks
Scanned originals with poor image quality remain the weakest input. So do agreements that were assembled from three precedents and never reconciled, which is common in long-lived supplier relationships.
We will revisit this once we have a larger sample. The current numbers are directional rather than settled.
Working notes from the Lawrs team. General information about legal technology and practice, not legal advice.