It's soon going to be less about having AI than where your AI lives.
As we close in on the second anniversary of ChatGPT's release, the generative AI conversation -- at least when it comes to the legal industry -- is starting to shift. After a year or so of fretting about generative AI hallucinating like it dropped bad acid and trying to figure out newer and better legal offerings, autumn has ushered a new narrative as the AI discourse matures and providers focus on taking generative AI and improving the whole experience of using the tool.
This emphasis on practical integration and respecting everyday legal workflows dominating this week's NetDocuments Inspire Conference. "It's not about bringing content to your AI, but about bringing AI to your content," CEO Josh Baxter explained as the show kicked off. The era of lawyers taking material and feeding it into external AI-driven applications has to end.
"What does it look like if you have to take your content to AI. It looks like you're starting to stand up a bunch of point solutions to solve each individual problem. And its not a scalable approach. It created governance issues and security risks for your organization, it creates an environment where you're dealing with multiple vendors your costs start to pile up, the workflow and the experience for your end users is frustrating and inefficient."
An analogy that kept coming up is the increasingly disjointed experience with streaming services with users hunting and pecking between apps searching for a single show. But TVs that have indexed everything and can jump to the right service for the show when asked make the user's experience better.
The company's guiding mantra to build an "intelligent document management system" envisions a bunch of features, but the most straightforward is the most significant. Taking a firm's corpus of documents and employing semantic search to deliver the material a lawyer needs, not necessarily the material that they can pick the right keywords to capture. With an auto-profiling capability, Chief Product Officer Dan Hauck explained, NetDocs will be able to vectorize documents in real time to empower these common sense searches. So when a lawyer takes a new, say, deposition transcript dropped into the system and enriches the document so it doesn't get missed because someone forgot to tag it appropriately at the time or crafted the wrong query down the road.
Earlier this year, in the eDiscovery context, I mused that generative AI's most powerful application might be as a user experience tool. Lawyers being able to type out natural, common sense queries and the system fill in all those nuanced gaps that seniors depend (successfully or not) upon juniors to figure out. Applying this to the document management context opens the door for a more simple and meaningful interrogation of a firm's stored knowledge.
NetDocuments also introduced the ndMAX Legal AI Assistant, promising to provide answers and insights directly within the document management system. This tool, currently in beta, will be available by the end of the year. Issues List Generator automatically compiles key issues from contracts or case files, allowing for a more focused review. The transactional side isn't left out of the fun, with the Contract Playbook Generator creating customized negotiation playbooks complete with clause suggestions and fallback positions, while the Due Diligence app automates document review in mergers and acquisitions, flagging risks for thorough analysis.
This is all on top of the apps that users are crafting themselves inside the system. For those, NetDocuments announced a new import/export feature allowing apps built inside one repository to be moved to another easily without requiring a new build.
All of which is to reiterate that the narrative around legal industry AI has moved toward improving the user experience, embedding AI in ways that deliver meaningful, practical results. An enabler of efficiency and accuracy, without requiring a steep learning curve or technological juggling -- perhaps without even requiring the user to realize that they're using AI to get these answers.