246,000 documents. Six days. A discrimination pattern, surfaced sentence by sentence.

The firm represented the plaintiffs in a fair housing case against a city council and its individual members, accused of blocking a proposed housing development through zoning denials, procedural delays, and pretextual objections motivated by the race and family composition of the residents the development was expected to serve. Discovery produced 246,000 documents — emails, internal city memos, text messages, and constituent communications — covering a three-year window across council members and city staff.

Defense counsel had been reviewing for weeks. Manual review of the full corpus would take a 25-attorney team approximately six weeks at standard rates. The litigation team had one week before the first depositions of council members and senior city staff. Anything missed would mean walking into a deposition without the document the witness was about to be asked to authenticate.

The firm pointed HarmCheck Rapid Deploy at the document corpus on Sunday evening. By Tuesday morning, the system had completed a full pass across all 246,000 documents — surfacing communications referencing protected classes, coded references to “neighborhood character” and “the wrong kind of resident,” and the pretext patterns documented in prior fair housing matters against municipal defendants.

Every hit was returned with the original document, the precise sentence triggering the classification, the category, and a confidence score. Senior associates worked from a curated subset of ~4,200 documents — a 98.3% reduction in manual review burden — backed by a complete audit trail of what had been reviewed, what had been excluded, and why.

By the end of day six, the litigation team had a curated set of ~4,200 flagged statements to review — each tied back to the source document, the precise triggering sentence, and the category that surfaced it. The judgment about what mattered, what to use, and how to use it stayed with the attorneys.

Instead of a 25-attorney team spending six weeks combing through 246,000 documents at standard rates, the litigation team spent six days reviewing a focused subset — with citations, source links, and category labels already in front of them. HarmCheck didn’t write the deposition outline, develop the case theory, or decide what to put in front of a witness. It just made sure the right statements were on the table in time for the team to do that work.