Canadian legal AI pioneer, Kira Systems, has released a major upgrade to its machine learning software that will allow better analysis of data stored in tables and forms.
The upgraded algorithms use layout and location information as well as content to help Kira perform analysis of structured data better, said the company. The move updates Quick Study, that it rolled out in beta to a few enterprises in summer 2014, then rolled out more broadly in winter 2015.
Noah Waisberg, CEO and co-Founder of Kira, explained that the benefits of this capability is that not all important material that clients had wanted to examine using AI-driven software was in a text document format. In some cases key data was held in tables, for example a table listing financial data that was key to the wider analysis the client was conducting.
Waisberg added that the improved capability would also handle forms, for example land registry documents or tax forms. It will do this without the need for special customisation or providing a law firm client with ‘consulting’ support from the AI company to create a one-off solution.
‘Quick Study was originally built to extract data from contract text. We found that a number of our users were happy enough with Quick Study that they were interested in using Quick Study to extract more structured information too. Quick Study 2.0 significantly improves the system’s performance on structured data,’ said Waisberg.
The continued development of Quick Study suggests a continuing demand from law firm clients for a ‘user friendly’ interface that can handle all their needs without complex training of law firm staff. To some extent clients can simply ‘plug and play’ and make use of the Kira system without special knowledge. This allows more lawyers to make use of it.
‘We continuously improve Kira’s technology, but never at the expense of the user experience. Maintaining an intuitive design is always of top importance,’ added Alexander Hudek, CTO and co-founder of Kira Systems, in a company statement.
Be the first to comment