From 2 Months to 3 Weeks: How a Pharma Company Cut 90% of Review Headcount with JND and Relativity aiR for Review

How did they do it?

• Developed and validated prompts with 90%+ recall before running aiR for Review across the full population
• Leveraged aiR for Review to surface approximately 50 highly relevant documents that traditional methods would have missed
• Eliminated the need for a large contract reviewer team, reducing cost and improving coding consistency

The Challenge: A Matter That Couldn’t Wait

When a major pharmaceutical company received a regulatory request involving several terabytes of data, the deadline left little room for error. Millions of documents needed to be collected, processed, reviewed, and produced within weeks – and the review itself needed to be accurate enough to stand up to regulatory scrutiny.

Traditional approaches presented real limitations. Keyword search routinely missed nuanced, context-dependent content. Active learning models needed seed examples that did not yet exist. And staffing a large contract reviewer team under a short and strict deadline risked inconsistency and unreviewed document exposure.

The pharmaceutical company turned to JND for help.

The Solution: Putting Relativity aiR for Review to Work

JND hit the ground running, proposing Relativity aiR for Review as the primary review engine for a few simple reasons: it could meet the deadline, hit the accuracy bar, and do it at a fraction of the traditional cost. With the client on board, they got to work.

The process began with a structured kickoff to define the key issues. From there, the team used conceptual clustering and attorney-guided sampling to build a representative document set for prompt development. Working collaborative with attorneys, they iterated on the prompt – comparing aiR for Review’s coding against attorney judgement and refining the instructions until AI and human coding reached high agreement. Then, they triggered a full-run validation.

JND also configured aiR for Review for PII/PHI screening – a use case where keyword search routinely breaks down – flagging sensitive content for attorney review within the same workflow. Reviewing document-by-document against a detailed instruction set, aiR surfaced approximately 50 highly relevant documents that no other method would have found.

"aiR for Review let us focus attorney time on case strategy, not document triage – and gave us confidence that nothing slipped through. We closed a matter that simply would not have been possible under this budget and timeline with any other approach."
Rachel Koy, JND Legal Administration

The Impact: Confidence, Delivered within 1 Month

The matter closed on time with 90%+ recall validated, zero unreviewed document risk, and a final production set exceeding 50,000 documents, all inside a single month. Eliminating the large contract review team reduced cost materially while improving coding consistency.

The client has since returned, engaging JND and aiR for Review for future regulatory requests, a pattern that reflects life sciences and pharmaceutical's ongoing reality. New clients now approach JND specifically requesting Relativity aiR because they've seen what it makes possible: matters completed on time that budget or deadline constraints would otherwise have put out of reach.

JND’s work on this matter demonstrates what’s possible when experienced practitioners combine the right methodology with the right technology. For JND, aiR for Review isn’t just a convenient tool – it’s crucial for bringing large-scale review capacity to matters that would otherwise be out of reach and taking on matters that budget or timeline constraints would otherwise make impossible.