A large city government had ~1.5MM records collected through targeted search based on search terms provided by opposing. Initial estimate for review was perceived as too high, and they planned to review in-house until an order from the court accelerated the timeline.
The review needed to be completed in under three weeks with 700k+ documents identified after de-duplication as hitting on search terms.
Our GenAI SOLUTION
Our Consultants dug into search terms and identified syntax errors in the original terms that led to over-inclusion. We further quickly identified a set of documents related to the core issues and moved them into our GenAI solution for deeper investigation.
Those core documents were used to enhance search terms, consolidate the requests for production, and gain key insights into the data early in the process. We joined a Meet and Confer with opposing, successfully proposing alternative search terms that reduced volume, then iteratively reduced the set removing false hits.
From there, data was pushed through another GenAI solution which identified obvious NR documents while our review team began training a TAR model. AI was used to code predicted Responsive families as well as a first pass of Privilege coding.
The full package of tool application resulted in less manual review and streamlined rolling productions until the deadline which was met with ease.
The Cost Saving RESULTS
- Cut review set from 700K+ to under 300K using refined search terms and targeted filtering
- AI/TAR/Human hybrid workflow accelerated review and privilege coding
- In under 3 weeks completed ECA, review, production, and privilege log
- Saved over $400K in review costs through intelligent triage and automation
- Delivered consistent, defensible results repeatable across supplemental collections, strengthening the City’s position with opposing counsel and the court
