Lighthouse, a leader in technology-enabled eDiscovery, compliance, and information governance services, announced Lighthouse AI for Review, a suite of review offerings targeting high-impact use cases commonly seen in large-volume, complex, time-sensitive matters. Lighthouse AI for Review is powered by a proprietary combination of AI and linguistic modeling—predictive AI for classification, generative AI to improve quality and accelerate downstream review, and linguistics to improve accuracy above what the AI is capable of alone.
“While much attention has been rightly paid to the impact of GenAI recently, the power and relevancy of predictive AI cannot be overlooked. They do different things, and there is often real value in combining them to handle different elements in the same workflow”
An early pioneer in deploying AI, since its inception in 2019 Lighthouse AI has been used to help legal teams review and analyze 3 billion documents across thousands of matters accurately, rapidly, and defensively.
“While much attention has been rightly paid to the impact of GenAI recently, the power and relevancy of predictive AI cannot be overlooked. They do different things, and there is often real value in combining them to handle different elements in the same workflow,” says Fernando Delgado, Sr. Director, AI & Analytics at Lighthouse. “We are the only company to have mastered fine-tuning predictive LLMs for eDiscovery, and it has saved our clients countless hours and dollars.”
Adding generative AI to that foundation has allowed Lighthouse’s experts to quickly spin up new solutions to solve specific client problems within the review process. “What sets this AI solution apart is the ability to truly collaborate on innovation,” says Christian J. Mahoney, Counsel and Global Head of e-Discovery and Litigation Technology at Cleary Gottlieb Steen & Hamilton LLP. “Rather than a one-size-fits-all approach, we’ve been able to shape the technology to fit our unique needs—turning our ideas into real, impactful solutions.”
The initial release of Lighthouse AI for Review includes the following use cases:
- AI for Responsive Review – As data volumes grow, scalable review is essential. Lighthouse AI for Responsive Review combines predictive AI with supervised fine-tuning to classify responsiveness while using Gen AI-driven relevance summaries and thematic coding to streamline review and QC. This approach reduces document volumes by up to 40% without sacrificing completeness. Lighthouse’s LLM-driven models require less training than traditional predictive models and provide greater consistency and an ability to adapt to scope changes than purely generative approaches, ensuring faster, more defensible results.
- AI for Privilege Review – Privilege review remains one of the most complex and time-consuming aspects of eDiscovery, a reality only exacerbated by data volume growth. By blending fine-tuned LLMs with matter-specific communication analysis, Lighthouse AI for privilege review is 60% more accurate than keyword-based models. Unlike other industry standard AI, the solution does not require manual administrative set up, resulting in workflows that are faster and more effective at the largest scale. For privilege logs, generative AI creates detailed, defensible descriptions that go beyond standard drop-down fields, reducing manual effort while improving clarity.
- AI for Image Analysis – In keeping with data trends, the volume of image-based data is growing rapidly, and there are no good options for handling this data at scale. AI for Image Analysis uses both predictive and GenAI to analyze image content, generate text descriptions, and cluster the descriptions in an easy-to-navigate UI designed to streamline review. This “human-in-the-loop” workflow enables legal teams to rapidly assess millions of images and identify those that require special handling.
- AI for PII/PHI/PCI Identification – Today’s documents are becoming more complex, making it harder to find sensitive information. Predictive AI with linguistic-based fine-tuning pinpoints these sensitive data types while avoiding the over-inclusion of irrelevant documents. For PII requiring entity linking, GenAI automates relationship mapping, replacing traditionally tedious, error-prone manual review.
“At Lighthouse, we launched our first LLM-based eDiscovery tool in 2019, and we have been tirelessly developing on that purpose-built foundation ever since,” says Ron Markezich, CEO at Lighthouse. “These new offerings are highly differentiated and designed to provide the most impact for the volume, velocity, and complexity of eDiscovery.”