BRG Cuts First-Level Review from 30,000 Documents Down to Fewer Than 500 with aiR Review

How did they do it?

• Used aiR Review to reduce first-level review on a complex, multi-language arbitration by 98%
• Prioritised and identified relevant documents beyond what keyword searching and TAR could surface
• Gave counsel a focused set of documents to work from, well ahead of the disclosure deadline

A Complex Dispute with a Fixed Deadline

BRG supported a multi-jurisdictional arbitration, brought by a state party against a consortium of international energy companies. The matter included approximately 1.5 million highly technical documents spread across several languages and jurisdictions, including the UK, the US, Southeast Asia, and Eastern Europe. Multiple teams worked through them at once, including cost recovery, environmental and corruption assessors, investigating project delays, corruption, and environmental compliance.

First-level review had to move quickly and be scoped in proportion to the claims in dispute, all while staying consistent and defensible. Dispute volumes can often grow significantly, but timeframes and disclosure obligations rarely adjust.

The usual tools weren’t getting the team what they needed. The data was low richness, so keyword searching struggled to surface the right documents and TAR was unlikely to find them quickly. Previous manual review had returned few results.

The team needed a faster, defensible way to find the documents that mattered across languages, without waiting on a lengthy manual effort. So, they turned to aiR Review.

The Solution: Prioritised, Multilingual First-Pass Review

With aiR Review, the BRG team could prioritise and identify relevant documents in first-level review, helping reviewers surface critical material for disclosure more efficiently.

aiR Review performed strongly across European and Cyrillic languages, and its document summaries helped the cost recovery, environmental, and corruption teams overcome language barriers and quickly assess relevance to their specific issues.

The Results: Faster Review, Earlier Insight, Greater Confidence

BRG completed the first-level review within three days, reducing the review population by about 98%. Because the documents were so highly technical, review on this matter had historically averaged around 20 to 25 documents per hour. At that rate, manually reviewing 30,000 documents would have taken an estimated 1,200 to 1,500 lawyer hours.

Beyond time saved, the team surfaced key documents and themes early, allowing them to focus on analysis instead of sifting through data. aiR Review also gave them a consistent, defensible process and a predictable review cost that counsel could take to their client for approval.

The legal team plans to use aiR Review across other teams and later tranches of data and sees potential to extend it into issue-based categorisation. They also see opportunity in future APAC matters involving CJK languages, where keyword search and TAR have historically struggled.

"aiR Review fundamentally changed the economics and speed of document review. On a complex, multi-jurisdictional matter, it enabled us to reduce a review population of approximately 30,000 documents to fewer than 500 for disclosure, while surfacing critical evidence that was difficult to identify through traditional keyword searching and TAR."
Erick Gunawan, Managing Director, BRG