Join our daily and weekly newsletters for the most recent updates and specific content of the industry AI's business. learn more
When a legalized research company Lexisnexis He created his AI assistant, so that he would want to issue the best way to reduce her knowledge of his knowledge.
Protégiver aims to help lawyers, his companions and a paramelery writing and make sure anything that comes in Complaints and words are correct. However, Lexisnexis did not want legal legal legal person; They wanted to build a man who learned a company company and is more social.
Lexisnexis saw the opportunity to introduce major language modules from proof and Missisnexis and professional tax.
“We use the best model for the particular issue as part of our multiple model procedure. We use their best response time,” Ricehl. “For using some cases, that will be a small language model such as a confidence or we perform the distilling to improve performance and reduce costs.”
While LLMS still applies in the construction of AI applications, some organizations to use small language modules (slices) or disturbing to become small versions of the same module.
Distilling, where a lmm “teaches” a smaller model, have been Be a popular way for many groups.
Small models work best for apps such as chatting or finished code, that was the lexisnexis for a protégé.
This is not the first time Lexisnexis raised AI clares AI, even before he puts the legal power name Lexisnexis + Ai in July 2024.
“We have used a lot of AI in the past, more around natural processing,” said Rhl. “That was a great deal in November 2022 when Chatgpt was launched, for before the abilities of conversation came behind, we were very interesting.”
Small Models, pulling and model route
Rihl Lexlxis told us using a variety of models from the majority of major model providers when they raised her platforms. Traditional procedures used for AI from Tarritical Models, Openi's GTTRA models
This method of enclosure methodo will help to break down each action usage for play on stage. To do this, Lexisnexis was to the architectural architect to turn between modules.
“We are violating whatever work it was played in a separate part to support the idea,” Raehl said.
For a Protégé, the company wanted to answer moods and models of answering of having better for legal practice. It so turned to the RUHL sits “of modules of models of models of models, largely of lolms or creative models.
“You do not need GPT-4O evaluation to a question, so we use it to turn it for models,” he said.
When a user asked Protégé, first model is a good “to evaluate the question, the purpose of that model will be a library to carry out the task or other model of products.
Currently, Lexisnexis is usually responsible for an incorrect channel although Reshl said it was using a neost of Claude “.” A Lexisnexis interested in using other open models are especially since the company came out Convolved Pictures Last year. Lexisnexis is currently reviewing Openi's reasoning modules including O3 for the platforms.
RUHL said it may also look at Gemini modules from Google.
Lexisnexis is moving with all the existing platform with the graph of herself to enable properties of property later.
The legal lagor
Even before Lexisnexis examined the ability to chat work in the legal business. In 2017, the A Test company did ai assistant That would compete with Ross Chieton and protégé in the Lexisnexis stage + AI, which bring with AI-services of Lexisnexis.
Protégé helps assist law companies with actions tendered to make notes. He helps to write legal brief and objections are based in companies and data documents suggesting new ideas in ethics, summarizes complex legal values.
“We'll see Protégé as the first step in personalization and abilities of Altehl,” REEHHHHS. “Think about the different types of lawyers: M & A, supporters, this is a personal assistant to do, what do other lawyers do.”
Protég, now competes against legal monitoring and other technology platform. Roseson Rutters creating O1-mini-mini-mini-mini-detailed for his Cooologic Legal Help. Harvey, a He raised $ 300 million from investors include Lexisnexis, a law assistant is also.
Source link