Box announces content retrieval system based on Nvidia’s Nemotron

Box presented Box Content Retrieval, a feature that builds Nvidia’s Nemotron embedding system into the core of its product. The company says the technology can make queries faster, more accurate and less expensive by using metadata to return only the most relevant document chunks to agents, rather than having them read entire files.

Kari Briski, Nvidia’s vice president of generative AI software, told TechCrunch that she expects queries to be four times faster, with 20% fewer errors and 90% fewer tokens used. Box expects to make the feature available early next year to all Enterprise Advance customers. The system combines Nemotron’s open-weight models with Box’s existing metadata and permission controls; Box cites lawyers as one possible beneficiary, since the tool could make it easier to find related clauses and precedents even when wording varies.