Marcelo Tostes is one of Brazil’s foremost law firms. With more than 500 employees, they have a multidisciplinary, integrated team that focuses on corporate law and innovative businesses.
The firm has positioned themselves as a business who is “working hard to remove barriers and strengthen the connection between Law & Technology.” All of this adds up to a business that is able to offer a personalized service to their clients along with top-notch legal representation, and local and international coverage.
The number of lawsuits in Brazil, per 1,000 in habitants, is almost five times more than countries like Germany, Sweden, Austria, and Israel. And Brazil is not shy about the amount of money they spend on their judicial branch, the system is overloaded with pending cases, in the amount of over 80 million lawsuits.
One of the leading contributors to this incredible congestion is is the need to classify legal pleadings. This process is, generally, done manually by one or two paralegals who may take a full day to read and classify 600 legal pleadings. It is a time consuming and costly process that is prone to human error.
The firms that Marcelo Tostes was working with were classifying the legal texts themselves. What they sought was a solution that would allow them to perform all of this classification themselves, without waiting for other businesses to perform those tasks. The solution would also need to cut down on the element of human error in the process.
The firms that Marcelo Tostes was working with were classifying the legal texts themselves. What they sought was a solution that would allow them to perform all of this classification themselves, without waiting for other businesses to perform those tasks. The solution would also need to cut down on the element of human error in the process.
The firm uses a 3rd-party API to retrieve legal texts relevant to their clients. These texts are then sent to the clients of Marcelo Tostes, at which point their individual paralegal teams begin the process of reading and classifying the texts that concern them.
With Kepler, Marcelo Tostes continued to use the 3rd-party API to retrieve the legal texts. The information was then added to a CSV file which was uploaded to Kepler. From there, the AI workflows were able to classify up to 1,000 legal texts in just a matter of minutes using Kepler’s batch interference capabilities.
What Kepler did was use the text provided in each of the texts to predict which legal classification the suit would fall under. The AI was able to predict the correct classification with 90% accuracy
The legal texts, along with their respective classification were then served to their clients via another API with the information needed to proceed available to them immediately.
Since Marcelo Tostes began using Kepler to sort and classify the vital information needed for their business to proceed with their cases, they have been able to vastly improve that classification speed. The result has meant much less time to organize the tasks and prioritize which legal cases are pursued.
Since the start of the mandate, Stradigi AI has worked closely with the firm to optimize the process, better preparing the data before feeding it into Kepler’s AI. As a result of training with Stradigi AI, the team at Marcelo Tostes has become AI fluent, capable of building their own models in the platform.
We now find that when the data is optimized, these texts are misclassified only 3.88% of the time—an astounding result that has improved the ability of the firm to focus on the work that really matters to their business.
of documents processed in minutes
