Case Studies

The traditional approach to document review and fact-finding was wasting valuable time and resources for a telecommunications company and their outside counsel during a complex litigation. The usual search and review methods were also failing to surface the critical insights counsel needed to prepare their litigation strategy and minimize risk in a timely manner. Lighthouse experts stepped in and transformed the process by integrating AI and linguistic modeling to streamline review, reduce costs, and get critical insights into the hands of the case team, faster.The Problem: An Inefficient and Ineffective WorkflowThe contract review team had initially been tasked with analyzing and categorizing each document for various factors, including: Responsiveness 17 Issue Codes 4 Levels of Confidentiality Privilege Hot/Key StatusThe Result: Linear Review at a Snail’s Pace A sluggish and expensive review process that drained resources—while burying counsel under a mountain of redundant documents that delayed key decisions and increased the risk of unwanted surprises. ‍The Pivot: A Better Approach to Fact-Finding and Doc ReviewRather than continue with the traditional approach, Lighthouse experts built tailored AI classifiers to tackle specific review tasks—e.g., confidentiality, privilege, and identifying key documents related to specific issues. The goal of this modern, more strategic approach was to help counsel remove large swaths of documents from the review queue, speed up review on the documents that remained, and get critical insights into the hands of the case team faster.The Result: Accelerating Review with AI‍Lighthouse AI and linguistics accurately classified confidentiality and privilege for the vast majority of documents (see above) and enabled Lighthouse search experts to quickly identify 1.6K unique key documents across a variety of areas for the case team. For the 120K documents that remained in the review queue, the review team was able to double their review pace because they could focus solely on reviewing for responsiveness.Read on to learn the details about how Lighthouse experts used AI and linguistic modeling to tackle specific classifications and fact-finding tasks.Confidentiality Classifications All responsive documents needed to be classified into one of four distinct confidentiality levels: Outside Counsel Eyes Only (OCEO) Confidential – Restricted​ Confidential​ Not-Confidential​To ensure precision in the AI model, Lighthouse experts collaborated closely with outside counsel to define the specific criteria for each level. With this input, they built tailored linguistic classifiers to automate the confidentiality classification. Samples were sent to outside counsel to test and refine the classifier before Lighthouse deployed it on the remaining document population. The result: Outside counsel only needed to review a total of 630 documents to validate and refine the Lighthouse AI linguistic classifier. From there, the classifier was able to accurately determine the correct confidentiality level for all responsive documents within the 590K document population. Lighthouse experts then continued to deploy the classifier on all newly collected documents. Identifying Telecom Agreements To protect confidentiality, as well as mitigate risks to the company, it was critical to identify all instances of specific types of agreements at issue in the litigation. Unfortunately, there was no definitive list of the parties involved in those agreements. To tackle this challenge, Lighthouse search experts created advanced linguistic models specifically tailored to recognize the unique language patterns within the types of agreement at issue. The result: Lighthouse linguistic models identified all 15K+ agreements within the document population. Lighthouse experts de-duplicated the agreements before delivering the comprehensive set to the case team well before production deadlines. Key Document Identification and Trial Prep A dedicated Lighthouse search team used a combination of linguistic modeling, advanced search technology, and unique search expertise to find the critical documents the case team needed to see across a variety of areas.The result: Lighthouse’s small team of search experts identified 1.6K unique and critical documents and excerpts of important language buried within the large document tranche.‍They quickly provided these documents to the case team in small, curated deliveries for a variety of fact-finding and trial preparation needs, including: 8 essential key document topic areas Third-party production gap analysis Documents showing evidence of fraud and malfeasance Deposition preparation kits Ad-hoc case team search requests as the case evolved The speed at which Lighthouse experts were able to find critical information—in combination with the value and uniqueness of the information they found—ensured that the case team could make quicker, more informed decisions throughout the remainder of case and reduce the risk of wanted surprises.Efficiency, Accuracy, and Strategic Value with Lighthouse AI and Linguistic Modeling By deploying AI and linguistic modeling, Lighthouse not only enhanced the efficiency and accuracy of document review but also empowered the legal team to make more strategic decisions faster. This modern, data-driven approach resulted in significant cost savings, time reductions, and an improved case strategy, minimizing risks and accelerating the pace toward trial preparation.

Two Months to Tackle Three Million DocumentsA financial institution with an urgent matter had two months to review 3.6M documents (2.4TB of data).With that deadline, any time that reviewers spent on irrelevant documents or unnecessary tasks risked missing their deadline. So outside counsel called on Lighthouse to help efficiently review documents.AI and Experience Prove Up to the ChallengeUsing our AI-powered review solution, we devised an approach that coordinated key data reduction tactics, modern AI, and search expertise at different stages of review.Junk Removal and Deduplication Set the Stage We started by organizing the dataset with email and chat threading and removing 137K junk documents. Then we shrank the dataset further with our proprietary deduplication tool, which ensures all coding and redactions applied to one document automatically propagate to its duplicates. AI Model Removes 1.5 Million Nonresponsive Documents To build the responsive set, we used our AI algorithm, built with large language models for sophisticated text analysis. We trained the model on a subset of documents then applied it to all 2.2M TAR-eligible documents, including transcripts from chat platforms. The model identified 80% of the documents containing responsive information (recall) with 73% accuracy (precision). The final responsive set consisted of 650K family-inclusive documents—18% of the 3.6M starting corpus. AI Supports Privilege Detection, QC, and Descriptions Our AI Privilege Review solution supported reviewers in multiple ways.First, we used a predictive AI algorithm in conjunction with privilege search terms to identify and prioritize potentially privileged documents for review. During QC, we compared attorney coding decisions with the algorithm’s assessment and forwarded any discrepancies to outside counsel for final privilege calls. For documents coded as privileged, we used a proprietary generative AI model to draft 2.2K unique descriptions and a privilege log legend. After reviewing these, attorneys left nearly 1K descriptions unchanged and performed only light edits on the rest.Search Experts Surface the 300 Documents Most Important for Case Prep Alongside the production requirements for the Second Request, Lighthouse also supported the institution’s case strategy efforts. Each tranche of work was completed in 4 days and within an efficient budget requested by counsel, who was blown away by the team’s speed and accuracy. Using advanced search techniques and knowledge of legal linguistics, our experts delivered: 130 documents containing key facts and issues from the broader dataset, for early case analysis. 170 documents to prepare an executive for an upcoming deposition. Beating the Clock Without Sacrificing Cost or QualityWith Lighthouse Review—including the strategic use of state-of-the-art AI analytics—outside counsel completed production and privilege logging ahead of schedule. The financial institution met a tough deadline while controlling costs and achieving extraordinary accuracy at every stage.

Searching for Evidence in 8TB of Chat and Technical Data Senior executives at an information technology company suspected that former employees had utilized company resources and intellectual property when starting a rival company. To determine whether litigation was called for, executives needed to find the most relevant documents within 8TB of processed data. The data was extremely complex, dating back 6+ years and consisting mostly of Slack data and attachments including highly technical documents, applications, logs, and related system files—tallying over ten million files. The company engaged a senior partner at an AM50 law firm, who recommended using keyword search terms, filters, and targeted linear review to find the “smoking gun” documents—which was estimated to take several months. The company came to Lighthouse looking for a faster, more strategic search alternative for their investigation. Pinpointing Key Docs with Linguistic Analysis Two Lighthouse search and linguistics experts met with company executives to learn exactly what information they suspected the former employees had misappropriated. From there, our experts created linguistic-based search criteria that go well beyond keywords, taking into consideration the unique vocabulary and syntax of software engineers and developers, the conversational quirks of Slack and other chat-based communications, and the coded language used by people who are trying to get away with something. The team delivered documents in 2 batches, refining their search based on input from the executives—and resulting in only 39 files for the company to review. Getting Results—and a Start on Case Strategy—in Days In less than 10 days, 2 Lighthouse experts pierced the subterfuge in the employees’ chat messages to reveal patterns in their behavior and attempts to cover their tracks. In all, we found 39 documents representing possibly questionable conduct, which required only 141 hours of eyes-on review. In comparison, using conventional analytics would have identified 5-20% of the search population as key documents—up to 50K documents to review in this matter. So in the end, Lighthouse saved the company over 3 months and nearly $200K.Armed with knowledge of the key events, timelines, and context of conversations buried within the data, the company was primed to begin litigation efforts and had a team ramped up to perform additional searches when needed.Lighthouse KDI vs Linear Review

Firms Needed Fast Analysis of 25M Documents More than a dozen international law firms—including a Joint Defense Group (JDG) of 11 firms and several firms representing defendants outside the JDG—were engaged in a complex cluster of cases spanning over 30 US jurisdictions. The total document tranche included over 25M documents. The firms needed to find and understand the key players, timelines, and nuances involved in each litigation, while also preparing for hundreds of depositions, witness interviews, hearings, and trials scheduled across the litigation universe. However, traditional approaches to fact-finding and litigation (i.e., document review, keyword searches, etc.) were drowning case teams in extraneous and duplicative information. They came to Lighthouse looking for a strategic, unified approach to fact-finding, led by experts who could deliver the key documents, information, and details the case teams needed—and nothing more. Custom Workflows Power Consistency, Speed, and Efficiency Our experts started by creating a topic map across matters, which helped them quickly provide case teams with the core themes in each jurisdiction while reducing redundant search work. From there, as case strategy for each matter developed, the Lighthouse team drilled down into more nuanced fact-finding to help surface the documents case teams needed to learn the key details of each matter, through strategies like: State/Jurisdictional Overview Workflow – We used advanced search technology to target key documents in incoming productions and categorize them by jurisdiction, providing case teams with an immediate thematic overview of key facts and timelines. Re-Deployable Linguistic Model Workflow – Lighthouse linguists developed models based on intimate knowledge of the language used within the datasets, then deployed them within proprietary search technology to sort documents into tiers based on the likelihood that they contained key information. Deposition Kit Bundle Workflow – By bundling deposition kit requests from the same jurisdictions and departments together, we could search across smaller collections of documents and take a deponent-agnostic approach. Previously Delivered Name Hit Workflow – We provided case teams with documents from previously delivered results, giving them an advanced start on deposition preparation while further reducing duplicative searching. These repeatable workflows significantly reduced the volume of searching and coordination required across matters and enabled Lighthouse experts to quickly zero-in on the exact documents needed—without wasting counsels’ time with redundant and unimportant documents. Critical Docs Found and Delivered Across Dozens of Matters and Hundreds of Kits Over the course of two years, Lighthouse experts prepared dozens of case teams for complex litigation and handled a deluge of competing deadlines, priorities, and ad hoc requests (totaling as many as 70 requests at a time). For the Joint Defense Group, this meant: Over 1,150 deposition kits across 24 matters, encompassing 245K unique documents Over 100 state overviews across 21 different jurisdictions, encompassing 80K documents For law firms representing individual defendants, Lighthouse provided an additional:150 deposition kits, encompassing 13K documents 30 defensive overviews across 20 jurisdictions, encompassing 6K documents 1.3K documents in response to ad hoc requests and trial support Each delivery was limited to essential information—including key themes and players in every jurisdiction, potential gaps in productions, lists of hot/sensitive documents and potential deponents, and key strategy documents—and avoided redundant and unimportant documents. The combination of innovative workflows and cutting-edge technology enabled Lighthouse to keep our team small and consistent throughout the engagement, so the entire effort was achieved by a handful of Lighthouse experts with institutional knowledge of every matter. Since this engagement, we have used the same workflows for other clients facing complex Multidistrict Litigation (MDL)—making Lighthouse key document identification one of the most valuable and scalable litigation technology solutions on the market today.