Array Guides Law Firm Switch to Relativity aiR for Review Mid-Project -- Cutting Review by 75% and Saving £50K

How did they do it?

• Introduced aiR for Review midway through a linear review of ~100,000 documents when timeline pressures made a traditional approach untenable
• Developed and validated targeted AI prompts within just four days through rapid client feedback
• Reduced the active review population to ~20,000 highly relevant documents for human validation
• Completed first-pass review in one week instead of four, delivering approximately £50,000 in cost savings

The Challenge: A Traditional Review Falling Behind

A full-service London law firm was managing a matter involving a data set of approximately 100,000 documents requiring first-pass review.

The team initially relied on a conventional linear review approach. A group of 10 human reviewers began analysing the data set using standard workflows including keyword searching, file type filtering, image identification, email threading, and deduplication. Basic analytics were applied to prioritise documents, but the process still required reviewers to manually assess a large proportion of the data set.

Despite these efforts, the review quickly became difficult to sustain.

With approximately 83,000–84,000 documents still remaining, the projected timeline for first-pass review was around four weeks. Estimated costs were roughly £70,000, excluding the additional oversight that would likely be required from a review manager.

The process was also highly resource intensive. Reviewers were spending most of their time manually scanning documents rather than focusing on higher-value tasks such as issue analysis or privilege review. The workflow also lacked flexibility – making it difficult to adapt if new data appeared or deadlines accelerated.

Recognising the timeline was becoming unattainable, the firm approached Array for assistance.

The Approach: Introducing aiR for Review Mid-Project

Working with Array, the team explored using Relativity aiR for Review to accelerate the process while maintaining high recall of relevant documents.

As the firm had not previously used aiR, the rollout began with a small test batch to validate the results and build confidence before expanding across the full data set. Implementation was both straightforward and rapid, allowing the team to move quickly from testing to full deployment.

The client provided prompt feedback within hours, allowing the Array team to develop and refine prompts within just three to four days. Once the optimal prompt structure was established, aiR for Review was applied across the remaining document population.

The workflow also proved highly flexible. Any additional or unexpected documents could be processed immediately without requiring additional setup, allowing the team to maintain momentum and avoid delays.

By combining AI-driven analysis with targeted human validation, the review process shifted from broad manual scanning to focused evaluation of the most relevant material.

The Impact: Faster Review, Lower Cost, Higher Focus

The results delivered immediate efficiency gains.

• Prioritisation of Material: 75% reduction in the remaining review population
• Significant Financial Impact: £50,000 in estimated savings
• Drastic Acceleration: Project completed in 1 week, instead of 4

With a significantly smaller and more relevant document population, the review team was able to concentrate on higher value tasks like privilege review, validation, and quality control rather than spending time on broad first-pass scanning.

The ability to see clear explanations for why documents were identified as relevant also helped build trust in the AI output and gave the legal team confidence in defending the review process if required.

Why It Matters: A Flexible Model for Modern Review

For the firm, the project demonstrated how AI-assisted review can enhance traditional workflows without sacrificing defensibility.

Initial hesitation around AI quickly eased once the team tested a small batch and saw detailed explanations accompanying each classification decision. The technology not only accelerated the review process but also introduced a level of flexibility that will benefit future matters – particularly when handling unexpected data sources or tight timelines.

The project also highlights how full-service firms can now manage large-scale document reviews more efficiently by combining human expertise with AI-assisted workflows.

“I recommend aiR for Review because it gives our clients peace of mind. They can focus on the documents that truly matter, meet tight deadlines, and feel confident that nothing critical is being missed. It’s not just faster – it makes the review process more manageable and defensible.”
Steven Facer, President, Array UK