At Thursday’s developer conference, Box unveiled new features that incorporate Nvidia’s Nemotron embedded system into the core of the product.
This new feature, called Box Content Retrieval, enables agents to answer queries much more effectively by leveraging embedded metadata created by Nemotron.
Box says the result for users is faster, more accurate, and lower cost queries. Kari Briski, vice president of generative AI software at Nvidia, told TechCrunch that he expects queries to be answered four times faster, with 20% fewer errors and 90% fewer tokens used.
The system will be rolled out early next year and will be available to all Box Enterprise Advance customers.
“This dramatically increases the relevance of our search technology to agents,” said Box CEO Aaron Levie. “Instead of agents reading through each document, they can send back chunks of documents that are most relevant to the user’s query or the task the agent is performing.”
This feature is an implementation of Nvidia’s Nemotron Embed project, which provides an open weight embedding model to improve automated search systems. Big labs often overlooked models like this that use LLM to preprocess data and allow agents to use it better. This area has been the focus of Nvidia’s open source modeling project, and its recent Nemotron, released in July, tops the RTEB benchmark for measuring search quality.
Box said the system would be most useful when combined with active representation, citing the example of lawyers, who would ostensibly be able to find relevant provisions and case law more easily under the new system. Nemotron’s embeddings go beyond simple keywords and map underlying concepts, so search tools can uncover relevant legal themes even when different expressions and languages are used.
Although the Nemotron model is available for free, the system also relies heavily on Box’s existing metadata and permission controls, making it difficult to recreate from scratch. “We’re probably packing five different technologies into one simple API,” Levie told TechCrunch.
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
