Box taps Nvidia Nemotron for content retrieval system

Box is adding Nvidia’s Nemotron embedding system to its product in a new feature called Box Content Retrieval. Announced at Box’s developer conference on Thursday, the feature is designed to help AI agents retrieve relevant document content to answer queries.

Nvidia Vice President of Generative AI Software Kari Briski told TechCrunch she expects queries to be answered four times faster, with 20% fewer errors and 90% fewer tokens used. Box says the feature will be available to all Enterprise Advance customers, with a rollout planned for early next year.

Nemotron creates metadata that helps the system find relevant passages, rather than requiring an agent to read each document. Box CEO Aaron Levie said the system can return document chunks most relevant to a user’s query or an agent’s task. Box also says the embeddings can help surface related legal concepts when searches use different wording, such as when looking for clauses and precedents.

The feature implements Nvidia’s Nemotron Embed project, which offers open-weight embedding models for automated retrieval. Nvidia’s most recent Nemotron release, in July, ranked #1 on the RTEB benchmark, which measures retrieval quality. The feature also relies on Box’s existing metadata and permission controls. Box has not given a specific rollout date beyond early next year.