LexisNexis Launches Lexis+ AI, Generative AI for Lawyers Powered by OpenAI and Anthropic

The solutions top use cases are relevant well beyond the law

LexisNexis announced the general availability of its new generative AI solutions, Lexis+ AI. The company said it had previewed the solution with law firms, corporate legal departments, and U.S. courts. While it is technically general availability, interested users must sign up for a demo before accessing the solution.

Lexis+ AI is built around four use cases. According to the announcement:

Lexis+ AI features conversational search, intelligent legal drafting, insightful summarization, and document upload capabilities, all supported by state-of-the-art encryption and privacy technology to keep sensitive data secure:

• Conversational search simplifies complex and time-consuming legal research by enabling users to conversationally interact with Lexis+ AI, explore new insights, and ask for adjusted and refined output.

• Document drafting instantly produces legal arguments, contract clauses, and client communications from a simple user prompt.

• Summarization functionality delivers case summaries in seconds with more content and capabilities coming soon.

• Document upload capabilities enable users to rapidly analyze, summarize, and extract key insights from legal documents.

Unstructured Data Intensive Industries

These are key capabilities for any knowledge management function. Finding, creating, summarizing, and querying natural language information represent some of the most impactful capabilities that large language models (LLM) enable.

At one time, some industry observers and analysts were certain that law, healthcare, and finance would be among the last industries to adopt generative AI. They based their predictions solely on a perceived requirement for accuracy in these industries that LLMs could not meet. However, these hot takes failed to recognize the tremendous impact LLMs can have across use cases in industries that accumulate, create, and consult large volumes of unstructured data.

Law, healthcare, and finance are actually early adopters of generative AI precisely because they are industries that have so much unstructured, written information. While the accuracy problem is not solved, it can be mitigated, and there are many use cases where the value is high enough that even after adding a verification step to the activities, there is still tremendous benefit.

For its part, LexisNexis says Lexis+ AI, “minimizes the risk of invented content, or hallucinations, and checks all citations against Shepard’s to ensure citation validation. The solution also offers users the ability to input specific citations to verify accuracy and flag when a citation might be wrong. Customers can give instant feedback within the product to continually improve product performance, content relevance, and overall product accuracy.”

In other words, the solution is “grounded” for truthfulness against its curated dataset, and it conducts verification before delivering a result. It also provides citations to make verification easy for users and allows users to provide feedback when they believe information is incorrect. This means that there is either a learning system implemented or, more likely, a process to collect and analyze feedback that can be used to update data sources and employed for retraining the models.

Built on OpenAI and Anthropic

LexisNexis said the new solution is built upon a combination of OpenAI and Anthropic LLMs, hosted on Microsoft Azure and Amazon Bedrock (via AWS), respectively. No one will be surprised to see OpenAI referenced in this announcement. It is a de facto standard for LLM-enabled solutions and leads in both mindshare and market share today.

Anthropic is definitely included because of its unique 100,000 token context window. This means Anthropic can analyze 75,000 words of text in memory compared to somewhere between 6,000 and 25,000 for OpenAI models. The video above shows a use case where documents are uploaded. The demonstration shows Lexis+ AI summarizing the documents and answering questions about them. If you have a lot of documents, OpenAI’s context window will be too small to meet the need.

A bonus for including Anthropic may be a reduced reliance on OpenAI. This may provide LexisNexis with more flexibility if OpenAI or Anthropic fail to keep up with new features developed by other LLM providers or raise prices. Building the solution to support multiple LLM endpoints likely means they can more easily swap out either provider for an alternative. This does not mean it will be easy, but it will be easier with a more flexible architecture.

Legal Efficiency

LexisNexis is not the first company to provide LLM-enabled solutions for lawyers. OpenAI has invested Harvey, which positions itself as generative AI for elite law firms, though it does not exactly explain what that means. Harvey has announced a partnership with PwC and the law firm Allen & Overy. A&O said that its lawyers used Harvey for 40,000 daily queries during its trial period.

Further along the adoption path is Spellbook, which launched in November 2022. The company, formerly known as Rally Legal, says that over 1,000 legal teams use its generative AI solution that is customized on top of the GPT-3.5 and GPT-4 models from OpenAI. It has a broad feature set that is focused on drafting, editing, reviewing, and negotiating contracts. The company told Synthedia earlier this year that it received over 30,000 requests to use its software in the first six months after launch.

LexisNexis will further accelerate generative AI adoption across the legal profession. The company has thousands of customers across more than 150 countries. It also hosts over 2.5 petabytes of information, which the company says is 150 times larger than Wikipedia. Lexis+ AI will certainly put generative AI within easy reach of more lawyers. Importantly, the generative AI features are provided by a trusted brand and integrated into a solution that many lawyers use frequently.