Legora Partners with OpenAI to Automate Legal Work

The Swedish legaltech company integrates OpenAI models at the core of its legal workflows and organization. It highlights time reductions of 90% for a Fortune 100 client, a 95% decrease in internal time spent on NDAs, and 9.5 hours returned to lawyers each week. The roadmap is reviewed every two weeks as the models progress.

Quantifiable Gains on NDAs, Contracts, and Weekly Workload

Legora reports a 95% reduction in the time its legal team spends reviewing NDAs, dropping from about 20% to less than 1%. For a Fortune 100 client, the company cites a 90% reduction in contract processing time, over $20 million saved in external litigation and due diligence, as well as 9.5 hours reclaimed per lawyer each week. These results are presented as already achieved for this client and measurable internally within the scope of NDAs.

Adoption and Change Management in Client Organizations

According to Legora, when more than 75% of executives actively use the platform, median usage within the organization is three times higher. The recommended approach is to start with the problem to be solved, then identify where AI can enhance the workflow. The publisher emphasizes building a network of internal champions to convert experimentation into sustainable adoption, the necessity of being ready to rebuild as models evolve, and designing for trust by integrating assessments, testing, and transparency from the outset.

Legal Use Cases: From NDAs to Multi-Contract Due Diligence

Internally, Emma Kadri Bergström's team has built a manual from historical agreements to automate an initial pass on NDAs and flag out-of-scope clauses. The company indicates that before this automation, these reviews accounted for about 20% of the team's capacity, and now they weigh less than 1%. In merger and acquisition due diligence, agents initiate contract reviews and relay information, which is then validated by lawyers. On the client side, Alex Fortescue-Webb highlights Tabular Review, capable of analyzing thousands of contracts in minutes on multiple issues to extend the depth of review. Risks located outside the traditionally sampled subset can thus be highlighted. A Legora plugin for ChatGPT also allows access to these capabilities within conversations.

Product Architecture and Use of OpenAI at the Core of the Platform

OpenAI is a key provider of models for Legora, which relies on these models for essential workflows ranging from agentic work to large-scale document review. The publisher claims to quickly integrate new available capabilities to make them usable features for lawyers. The use of tools such as ChatGPT Work and Codex is also expanding internally, with OpenAI technology structuring both the product and the organization. Max Junestrand insists that the company remains AI-native and claims access to on-demand legal services. This positioning dates back to the early trials made possible by GPT-3.5 and aims for services executed on demand within lawyers' workflows.

Engineering Efficiency and Operational Responsiveness

Jacob Lauritzen explains that coding agents have accelerated the generation and selection of ideas by enabling the prototyping of numerous approaches in parallel, exploring up to 20 solutions in just a few days. These agents participate in diagnosing incidents by exploring millions of lines of logs and testing multiple hypotheses simultaneously, potentially leading to a fix in about five minutes after a report. Clients perceive the effects through faster resolutions, earlier feedback to product teams, and earlier testing, which, according to Lauritzen, allows for quicker feedback and product improvement.

Company Growth and Prospects for More Autonomous Agents

Over the past year, Legora has seen its number of offices grow from three to eighteen, successfully completed three funding rounds, and made five acquisitions, a pace that the legal team had to closely follow. According to Emma Kadri Bergström, artificial intelligence has taken on tasks that would otherwise have been assigned to external providers while enhancing quality. Looking ahead, Jacob Lauritzen anticipates agents capable of handling lengthy and contextual tasks, such as due diligence on millions of pages conducted overnight with analysis available by morning, as well as real-time collaboration between lawyers and AI made possible by faster models. He asserts that the human-AI mix yields the best results and notes that the roadmap is reviewed every two weeks based on model progress, supported by team requests. Emma Kadri Bergström emphasizes the challenge of avoiding the bottleneck image and becoming a business partner. If the company meets its goals, legal teams could take on previously impossible tasks and dedicate more time to high-value decisions. Max Junestrand summarizes this period as the best time to practice law. These ambitions are part of a strategy initiated in 2023, driven by its leadership, which focuses on an agentic operating system and the industrialization of on-demand legal tasks in a context of high time constraints.