Startup Luminance is set to launch an AI assistant for lawyers, designed to edit contracts and annotate non-disclosure agreements using a proprietary model trained on over 150 million legal documents.
The new tool, called Agent Lumi, will be available for a limited testing phase for customers by the end of the year. The legal AI market is highly competitive, with various companies aiming to help lawyers streamline the process of reading, amending, and negotiating business documents. For instance, Harvey, which provides generative AI solutions for lawyers and is supported by OpenAI, has gained significant attention, while Clio, a platform for document summarization and billing, recently completed a $900 million funding round in July.
Founded in 2015 by mathematicians from Cambridge, Luminance initially focused on automating contract reviews. Among its early investors was Mike Lynch, the former CEO of Autonomy, who was acquitted of fraud charges related to the company’s sale and tragically died earlier this year when his superyacht sank off the coast of Sicily.
Luminance’s first clients were law firms that allowed the company access to their legal contracts, aiding in the training of its AI. CEO Eleanor Lightbody emphasizes that Luminance’s specialized dataset, combined with tailored machine learning algorithms, differentiates its AI from others. Co-founder and chief technical architect Adam Guthrie notes that the model’s insight into a business’s contracts provides a comprehensive understanding of its operations.
Agent Lumi will compete with established AI agents, such as Spellbook’s Associate, which was recently launched as a comprehensive tool for legal work. Like Harvey and other AI applications, Spellbook is built on OpenAI’s foundational model.
The launch of Agent Lumi follows the introduction of Luminance’s chatbot, Ask Lumi, a generative AI model capable of answering business inquiries and revising contract language. Earlier this year, Luminance also released an auto-markup tool to review contracts and align clauses with company standards, although it operated in a more passive manner. In contrast, Agent Lumi is designed to be proactive, performing multiple tasks simultaneously through a “chain of reasoning.” Lightbody describes these tasks as repetitive and automated, enabling legal professionals to concentrate on higher-value strategic thinking.
Unlike more general AI models, Luminance highlights that its “legal-grade” AI is specifically trained to avoid errors, or “hallucinations.” Rather than providing incorrect information, the model is designed to refer to human professionals when uncertain.
This feature is particularly significant for lawyers, who require precise and accurate language. “If the AI doesn’t know the answer, it tells the human who does,” Lightbody explained. To build trust, users will have the ability to review every step taken by Agent Lumi, including the sources of information within contracts.
Luminance acknowledges that while its AI assistant offers advanced capabilities, it remains a tool for human lawyers, who are ultimately responsible for any inaccuracies.
Early partnerships with large law firms have been crucial for Luminance as it seeks to expand into new markets. Lightbody notes that if a law firm adopts their software, it often leads to adoption by the firm’s clients as well.
Luminance is also broadening its reach beyond the legal sector, increasingly targeting corporate clients. The UK-based company raised $40 million in a Series B funding round in April, though it did not disclose its valuation at that time. Additionally, Luminance has expanded into the U.S. market, opening offices in Dallas and increasing hiring on the West Coast, with more than a third of its annual revenue now generated in the U.S.
Lightbody asserts that scaling operations enhances product quality, as diverse data variations—not just the sheer volume of information—are key to effective AI technology.
