Box Brings Nvidia's Nemotron Embedding System to Its Core Product
At its developer conference on Thursday, Box unveiled a new feature that embeds Nvidia's Nemotron embedding system directly into the heart of its platform. The capability, called Box Content Retrieval, enables AI agents to answer queries far more effectively by leveraging the embedded metadata generated by Nemotron.
According to Box, the result for users is faster, more accurate, and less expensive queries. Kari Briski, Nvidia's VP of generative AI software, told TechCrunch that she expects queries to be answered four times faster, with 20% fewer errors and 90% fewer tokens used.
The system will be available to all Box Enterprise Advance customers, with rollout planned for early next year.
How It Works
"This dramatically boosts the relevance of our search technology for agents," said Box CEO Aaron Levie. "Instead of an agent having to read through each document, we can send back chunks from the documents that are most relevant to the user's query or whatever task the agent is doing."
The feature is an implementation of Nvidia's Nemotron Embed project, which offers open-weight embedding models designed to improve automated retrieval systems. Major labs have often overlooked such models, which use LLMs to pre-process data so it can be better utilized by agents. The sector has been a focus for Nvidia's open-source model efforts—its recent Nemotron release in July topped the RTEB benchmark, which measures retrieval quality.
Built for Active Agents
Box says the system delivers the most value when paired with active agents. The company cites lawyers as an example: under the new system, legal professionals would ostensibly be able to locate relevant clauses and precedents more easily. Because Nemotron's embeddings go beyond simple keyword matching to map underlying concepts, a search tool can surface related legal themes even when different wording or languages are used.
Hard to Replicate
While the Nemotron models are freely available, the system also relies heavily on Box's pre-existing metadata and permission controls—factors that would make it challenging to recreate from scratch. "We're probably packing in five different technologies into one simple API," Levie told TechCrunch.
via TechCrunch AI