Box revealed a new capability at its developer conference on Thursday that embeds Nvidia’s Nemotron system directly into its core product. The feature, named Box Content Retrieval, is designed to help AI agents answer user queries more effectively by relying on embedded metadata generated by Nemotron.
Box stated that the integration results in faster, more accurate, and less expensive queries. Kari Briski, Nvidia’s vice president of generative AI software, provided specific metrics to TechCrunch regarding the expected performance gains. She said queries should be answered four times faster, with 20% fewer errors and 90% fewer tokens used.
Box Content Retrieval and Agent Efficiency
The system will be available to all of Box’s Enterprise Advance customers, with a rollout scheduled for early next year. Box CEO Aaron Levie explained that the technology changes how agents interact with documents.
“This dramatically boosts both the relevance of our search technology for agents,” Levie said. “Instead of an agent having to kind of read through each document, we can send back chunks from the documents that are most relevant to the user’s query or whatever the task is that the agent is doing.”
Box Content Retrieval is an implementation of Nvidia’s Nemotron Embed project. This project provides open-weight embedding models intended to improve automated retrieval systems. Nvidia has focused on this sector with its open-source model project, and its recent Nemotron release in July topped the RTEB benchmark, which measures retrieval quality.
Legal Use Cases and Technical Complexity
Box indicated the system will be most useful when paired with active agents. The company cited lawyers as a primary example, noting they will be able to find relevant clauses and precedents more easily. Because Nemotron embeddings map underlying concepts rather than relying on simple keywords, the search tool can surface related legal themes even when different wording or language is used.
While the Nemotron models are freely available, the system relies heavily on Box’s pre-existing metadata and permission controls. Levie told TechCrunch that these existing structures make it difficult to recreate the system from scratch. “We’re probably packing in five different technologies into one simple API,” he said.
Source: TechCrunch
