Unnamed Entity

A global pharmaceutical company managing five concurrent litigation matters faced escalating legal costs driven by redundant document review. In large-scale pharmaceutical litigation, custodian overlap across matters is common — the same documents, email threads, and privilege determinations surface repeatedly across separate cases, yet traditional matter-siloed workflows force attorneys to re-review the same material from scratch each time. With no mechanism to carry forward prior coding decisions, the company was paying for duplicated attorney time, accepting inconsistent privilege and responsiveness determinations across matters, and forfeiting strategic insights that a connected view of its litigation portfolio would have revealed.

Lighthouse first optimized review for a single matter, deploying its AI-powered analytics to build a custom workflow that reduced data volume and improved classification accuracy using machine learning and predictive analytics. Based on those results, the company expanded the engagement to all five active matters. Lighthouse then connected the matters within its platform, enabling the AI to cross-reference documents at the point of ingestion for each new matter. Overlap detection ran at intake — before review began — so past attorney work product (privilege codings, redactions, responsiveness determinations) could be automatically applied to recurring documents. Classification models trained on prior matters informed each subsequent one, improving accuracy with every additional dataset ingested.

Across five connected matters, Lighthouse AI identified 105,300 overlapping documents, generating processing cost savings and early strategic intelligence at the outset of each new case. The system reused 26,450 privilege codings and 4,865 redactions, directly eliminating redundant attorney review hours. Beyond volume metrics, the AI surfaced custodian- and domain-level patterns — one outside counsel's email domain was coded privileged over 95% of the time; one custodian had 20,000 documents collected across matters but only 10 ever reviewed. These insights enabled counsel to make faster, more consistent decisions from day one of each new matter.

• 105,300 overlapping documents identified across 5 matters
• 26,450 privilege codings reused
• 4,865 redactions reused
• ROI compounds as each new matter refines classification models

• Cross-matter AI delivers compounding returns: connecting matters unlocks efficiency gains that are structurally impossible in a single-matter workflow, as classification models improve with each additional dataset.
• Privilege and coding consistency is a risk issue, not just a cost issue — reusing attorney work product across matters reduces the risk of conflicting determinations that opposing counsel can exploit.
• Overlap detection at intake, before review begins, is the highest-leverage intervention — it shapes resource allocation and case strategy from day one.
• Custodian-level analytics can reveal systemic over-collection; knowing which custodians produce near-zero reviewed documents allows earlier scope decisions on future matters.