Due diligence and discovery share the same tedious pattern: open a document, answer the same set of questions, close the document, repeat. Tabular Review was built to collapse that loop. Instead of reviewing one document at a time, lawyers define the questions once and let the Council answer them across the entire document set.
A Playbook is a schema
In CouncilGPT, a Playbook is a set of columns. Each column is a question the lawyer wants answered: What is the governing law? Is there a change-of-control clause? What is the liability cap? Is there a non-compete? The Playbook can be saved, versioned, and reused across matters. It turns implicit review standards into explicit, shareable instructions.
Once the Playbook is attached to a matter, the Council reads every document in the vault in parallel. For each document, it extracts the answer to each column and records the source clause. The result is a table: one row per document, one column per question.
Every cell is a link
The table is not a static export. Each cell carries a citation back to the original document and the specific clause that produced the answer. A reviewer can click any cell, open the file at the relevant passage, and verify the extraction. When a cell looks unusual, the system flags it as an outlier against the rest of the set.
Living tables
Documents added late in the deal do not require a manual refresh. The table updates itself. Finished tables can be exported to Excel for client distribution or turned into a summary memorandum. Tabular Review changes the unit of work from the document to the dataset, and that is what makes large-scale review manageable.