Clayton Utz has built a suite of tools to automate general and specialised legal tasks with OpenAI and RelativityOne, and is looking to save its lawyers more time by trialling Lexis+AI to generate documents.
Source: Clayton Utz
The law firm, one of Australia’s ‘big six,’ will evaluate Lexis+AI’s ability to generate first drafts of documents including advice to clients, internal emails and court filings such as statements of claim.
After LexisNexis announced Clayton Utz’s participation in an Australian preview of the solution, which is only commercially available in the US, the firm’s chief executive partner Emma Covacevich said: “Lexis+ AI will allow us to make even more effective use of legal information resources.”
The Lexis+AI pilot follows a series of previous investments in automation.
Clayton Utz has a team of over 100 staff dedicated to building tools for both firm-wide legal tasks and niche operations specific to environmental, social and corporate governance (ESG), employment, property and other fields of law.
Data broker LexisNexis, most known for its legal document subscription services, tapped Clayton Utz and Holding Redlich to provide feedback on Lexis+AI’s ability to automate tasks within the Australian legal sector before its commercial release in the Asia Pacific in the coming months.
“As LexisNexis is a key content provider, we see its AI solutions having an important role in our own AI program and helping us to deliver a new level of client experience,” Covacevich added.
Built on two LLMs — Anthropic’s Claude 2 and OpenAI’s GPT-4 — and trained on LexisNexis’ 1.23 million court opinions, statutes, filings, and secondary materials, Lexis+ AI automatically drafts documents in response to lawyers’ prompts.
LexisNexis granted Clayton Utz preview access because it has already rolled out Lexis+AI’s more research-focused automation tool Lexis+, to its roughly 1500 staff, including just under 200 partners, a Clayton Utz spokesperson told iTnews.
“Lexis+ is providing our solicitors a superior platform that is easier and faster to use — research takes less time due to the use of extractive AI. We are truly excited to see what the generative AI [Lexis+AI] solution holds for us.”
The automation use cases that Clayton Utz will trial with Lexis+AI, and has already rolled out with Lexis+ and the tools that it’s built with RelativityOne, Microsoft Azure OpenAI Service and other SaaS partners, include both automated research assistance and document-generation functions; Lexis+AI will advance its use of the latter.
Document generation: pros, risks and safeguards
Trialling Lexis+AI will not be the first time that the firm has sped up its workflow by automating content production; some of its in-house tools have document-generation functions that have “freed up our lawyers’ time significantly” Clayton Utz partner William Howe said last year.
Howe heads the firm’s data analytics practice within its forensic and technology services group: the engineers of tools like AcuiRE, which produces some of the reports that lawyers need to provide to clients undertaking compulsory acquisition of land parcels across multiple regions for large government infrastructure projects.
The RelativityOne-based solution “automates the task of collating documents for review including reporting, notifications to parties and generating output documents saving hundreds of hours of manual time.”
Clayton Utz’s internal AI working group evaluates both the time efficiencies won through automating tasks like writing documents and emails with platforms like Lexis+AI and the risks of outsourcing important tasks to technology.
A LexisNexis statement sent to iTnews said that the company “anticipates” that Lexis+AI will save “Australian lawyers 11 hours per week,” but lawyers are also weary of the career-shattering risks of LLMs producing documents with hallucinated information.
Last year a US-based lawyer was charged with “acts of conscious avoidance and [providing] false and misleading statements to the court” for filling a ChatGPT-written submission that referenced precedents it made up.
LexisNexis Asia Pacific managing director Greg Dickason told iTnews that to prevent references to hallucinated statutes or case law, LexisAI+ checks the citations in documents it outputs against Shepard’s Citations and the vendor’s other proprietary citation indexes.
Clayton Utz’s internal AI working group’s legal risk frameworks also assess the potential for inhouse and acquired generative AI solutions to violate “confidentiality” and “intellectual property,” laws, which Dickason said Lexis+AI also has safeguards against.
“Prompts are encrypted and the model does not learn from them. They are also deleted so IP is protected,” he said.
Lexis+AI is also said to safeguard against the risk of document-generation that would cause lawyers to get into the habit of producing substandard work.
After producing the document that the lawyer has prompted, Lexis+AI encourages them to review the output and use additional prompts to tweak its language to the intended tone.
Automated research assistance
“Clayton Utz’s entire team is using Lexis+,” which it adopted in July, the spokesperson said.
The firm made Lexis+’s “Argument Analyser” feature “available for everyone to use in the firm,” the spokesperson added.
Argument Analyser reads over lawyers’ documents and then recommends ways to improve their litigation strategy and also suggests relevant case law that they should reference in passages.
“Our lawyers use it for reading generally into a topic based on the text entered; searching multiple cases and then finding associated cases in common and to check to make sure they have included all relevant cases before sending the advice to clients.”
Lexis+’s other research assistance capabilities include automated comparisons of current and previous legislation to highlight changes over time and a feature that responds to users’ natural language search queries about case law to find answers more efficiently.
“We saw the opportunity that Lexis+ would bring; it allows our solicitors to be more strategic and deliver better outcomes to our clients.”
Tailoring tools to legal fields
Last year the firm partnered with Microsoft Azure OpenAI Service and developed an ESG tool that extracted key points from a backlog of climate change case law, then used ChatGPT-4 to succintly articulate and structure it.
Clayton Utz forensic and technology services director William Howe told Lawyers Weekly that the output still needs to be checked for accuracy and completeness, but incorporating automation made reviewing and summarising the cases months faster.
“The use of this technology is saving us time,” Claire Smith, the head of Clayton Utz national environment and sustainable development practice, added.
Howe’s team also “enhanced” a tool developed for Clayton Utz’s industrial relations lawyers last year, The Lawyer Mag reported.
AcCUrate, launched in 2021, fetches data from payroll systems and other sources to audit organisations’ compliance with employment laws.
“Our ‘AcCUrate’ tool [is] used to help clients validate and remediate employee payment issues, with AI that processes industrial instruments and generates validation code automatically and reviews for issues including wage compliance [and] unreasonable overtime risks,” Howe said in June.
The legal sector’s race for AI
The legal sector is among industries betting the most on generative AI and other emerging technologies for competitive automation solutions.
Allens, Lander & Rogers and MinsterEllison have found several use cases for Microsoft Copilot since its launch in February last year, for example.
Lander & Rogers also launched an AI Lab earlier this month to prototype customised tools, and last year it developed an automated privacy impact assessment tool.