LexisNexis, the legal research giant, has decided that bigger isn’t always better. Their new AI assistant, Protégé, is proof that sometimes, less is more—especially when it comes to billable hours. 🕴️
Small Models, Big Ambitions
Protégé isn’t your run-of-the-mill ChatGPT wannabe. Instead of throwing a bloated LLM at every legal problem, LexisNexis opted for a leaner, meaner approach. They’re using small language models (SLMs) and distilled versions of larger ones to handle tasks like drafting legal briefs, proofing documents, and ensuring citations are airtight. Jeff Reihl, LexisNexis’ CTO, summed it up perfectly: “You don’t need GPT-4o to assess a query.” Translation: Why waste a sledgehammer on a thumbtack?
The Fine-Tuned Future
Protégé’s secret sauce? Fine-tuned models. LexisNexis uses Mistral for query assessment and switches to more sophisticated models like Claude or GPT for heavier lifting. It’s like having a legal intern who knows when to ask for help. And let’s not forget the RAG (Retrieval Augmented Generation) capabilities powered by LexisNexis’ own knowledge graph. It’s not just regurgitating case law—it’s actually useful.
The Competition Heats Up
Protégé isn’t alone in the legal AI arena. Thomson Reuters has CoCounsel, and Harvey is out there raising $300 million like it’s Monopoly money. But LexisNexis is betting on personalization. Their vision? Every lawyer gets a bespoke AI assistant tailored to their specific practice area. Imagine that: an AI that knows your M&A from your IP. It’s like having a paralegal who doesn’t need coffee breaks. ☕ So, is Protégé the future of legal tech? Maybe. But one thing’s for sure: it’s a step toward AI that actually understands the job—not just the jargon.