The matter

A law firm instructed Jandr to manage disclosure for a commercial arbitration. We were brought in early, joining scoping consultations with the law firm and client before the issues for disclosure had been finalised between the parties.

The data

The client’s IT department conducted searches against centralised and cloud master stores, uploading 560,000 documents directly into Everlaw via its secure cloud connectors — no interim transfers, no media. Around 208,000 duplicates were removed on upload, leaving 352,000 documents for processing. We ran the agreed search terms through Everlaw and generated detailed hit reports in seconds. 68,500 documents were responsive and required review.

The challenge

First level human review of 68,500 documents would have cost around £285,000 using law firm fee earners, or around £91,000 via third party managed review — in both cases before law firm management overheads. Either way, around 160 review days, or over three weeks for a team of ten.

What we did

The law firm opted for AI first level review across seven distinct issues. We created a tailored prompt for each issue, sampled 200 documents per issue for the law firm to validate, refined the prompts, then ran them across the full dataset. First level review was complete in minutes. The AI platform charge was around £12,500, with law firm and Jandr oversight equivalent to that which would have accompanied a human review.

The outcome

5,800 documents were identified as relevant at first level. Second level review confirmed 76% of those as relevant. Sampling of the documents coded not relevant confirmed that no further action was required, thereby avoiding human review costs of 90% of the documents responsive to search terms. The client met its disclosure obligations on time, within budget, and with a defensible and documented methodology.