The Challenge

A leading construction firm faced a seven-day disclosure deadline to assess 6,000 documents for privilege and commercial sensitivity. The initial review appeared overly inclusive, risking wasted effort and increased costs. The goal was to validate the accuracy of the first-level review, minimise unnecessary second-level checks, and keep reviewers focused on privilege. By leveraging aiR for Review, Relativity’s GenAI-powered predictive coding tools, the team aimed to streamline quality control, maintain defensibility, and meet the disclosure deadline efficiently.

The Solution

To meet the demanding disclosure deadline and ensure accuracy, the team integrated aiR for review into their workflow as a second-level quality control tool - combining AI precision with expert oversight to accelerate review and maintain defensibility.

• Prompt Engineering: Transformed the client's review protocol into a natural-language prompt trained and refined through multiple test iterations.
  
  
• Data Culling: Used Relativity ECA tools to automatically remove 1,439 duplicate and near-duplicate documents, streamlining the data set. 
  
  
• AI-Driven QC: Deployed aiR for Review (powered by OpenAI's GPT-4 Omni) across 4,561 documents to validate first-pass coding decisions.
  
  
• Reviewer Focus: Human reviewers concentrated on privilege while aiR for Review handled relevance QC, saving significant time and cost.

The Results

• 8-hour aiR for Review run time

• 4,561 documents analysed

• 1,439 duplicates removed pre-review

• 90+ hours of manual review time saved

• 1,357 documents excluded from second-level review