LexisNexis has clarified how its new Protégé Legal AI Assistant connects to the holiest of holies, AKA your law firm’s Document Management System (DMS), to allay any fears around over-connectivity to legal secrets.
Last week the legal tech giant announced its wide-ranging Protégé genAI product, with a strong focus on personalisation and learning how each lawyer uses the tool (see article). But, due to potential concerns over how it may or may not connect to your Fort Knox of prior work product and related information, Jeff Pfeifer, Chief Product Officer in North America and UK, has sought to calm any nerves.
He said in a statement: ‘[Protégé] will replace the current AI Assistant on Lexis+ AI. Lexis+ AI subscribers will automatically inherit Protégé when it is commercially released later this year.
‘For those who don’t have DMS integration, users can manually upload documents to Protégé, ensuring that they have complete control over the data and information that the AI system can access and utilize for personalization.
‘For customers who want even greater personalization, our recent Henchman acquisition accelerates the ability to enrich data from their document management systems (DMS). Customers who select DMS services will be able to connect Protégé to their organization’s DMS. This will enable new capabilities, such as drafts in a specific tone, style, or format based on internal standards or individual writing styles.’
And it’s worth adding that one key aspect of Henchman has been that it can be tailored to strictly adhere to any firm’s particular security barriers and so it won’t drag up information from documents it’s not meant to be looking at.
Why This Is A Big Deal
Several times now the issue of connecting a broad AI system with multiple ‘skills’ to a law firm DMS has come up in interviews between Artificial Lawyer and law firms with new genAI products.
For example, Bird & Bird, which is using Leya doesn’t connect its DMS to the AI system and instead ‘ferries’ documents into it, as and when needed. This collation from the highly confidential motherload of legal data is done by the lawyers working on a new matter, who follow internal guidelines and only pick out what they’re allowed to tap before it’s pushed into the genAI system.
Meanwhile, CMS also noted that when they are using Harvey, another genAI system that can leverage a firm’s own work product, they also have not ‘hardwired’ it into their DMS (see article).
But, why does it matter? As explored by this site, central to the power and productivity of any genAI system in the legal world is its ability to tap quality data that’s relevant to a lawyer’s often highly specific needs.
Lawyers that leverage a genAI system to help them draft a document want the most up-to-date and relevant language carried up to them at the point where they are working. They don’t want generic legal language, but instead the wording and terms their own firm has used before.
But, what if some of this perfectly formed language you want to reach is associated with documents that can’t be tapped yet by everyone at the firm, perhaps because of commercial sensitivities, or a client has flat-out banned you from running an AI over their stuff?
Plus, even if results have all the key names redacted when such documents are analysed and key passages are surfaced to help draft a new contract related to a similar matter, some clients may still not be happy about being so transparent to a third-party AI.
Hence why LexisNexis has unusually issued a follow-up statement about how its Protégé Legal AI Assistant does or does not connect to your DMS.
Conclusion
Legal AI’s great opportunity is in being able to tap all of your data so insights and the best possible language can be surfaced for instant use, saving lawyers hours and perhaps even reducing risk for the clients.
Equally, legal AI’s great challenge is that the optimum collection of data is not always going to be easy to tap for every user. When it comes to case law, not every legal tech company can have access to the vast troves of court data some of the larger ones have. And, as explored above, just because you are a big firm with 100,000s of documents related to prior matters doesn’t mean each lawyer can tap all of them to improve their work.
That’s not to say AI systems should not tap your DMS. With the right guardrails in place and with your most sensitive files locked away out of sight from the AI, then there is arguably no reason not to let an LLM tap your work product (within the boundaries of your own business, naturally). After all, your bank vault of past deals is where AI will find the best examples to draw from.
Yet, lawyers – as part of a highly risk-averse profession – understandably don’t always want to cede their knowledge management powers to third-party software. Hence, as Pfeifer explained this is indeed all about control: ‘For those who don’t have DMS integration, users can manually upload documents to Protégé, ensuring that they have complete control over the data and information that the AI system can access and utilize for personalization.’
In short, there is a way forward that addresses law firm concerns.
—
(Pic: RT – the old bank vault at The Ned in London.)