Box has integrated the Nvidia Nemotron model into its content retrieval system

Box has announced the launch of a new content retrieval system built on agent-first technologies and the Nvidia Nemotron language model. The solution is focused on automating request processing within corporate data repositories.

According to the stated specifications, the new system demonstrates a significant performance increase compared to previous iterations. Requests are processed 4 times faster, while the error rate in responses has decreased by 20%. Algorithm optimization has reduced token consumption by 90%, which significantly improves the cost-efficiency of using the model at an enterprise scale.

The integration of Nemotron allows the system to act as an intelligent agent capable of more accurately interpreting the context of user queries and extracting relevant information from large volumes of documents. This solution is aimed at reducing the time employees spend searching for necessary data within the Box corporate environment.