Box integrates Nvidia Nemotron for faster agent queries
Box unveiled Content Retrieval at its developer conference, embedding Nvidia's Nemotron system into its core product. The feature enables agents to answer queries four times faster, with 20% fewer errors and 90% fewer tokens used, according to Nvidia VP Kari Briski. Rather than reading full documents, agents receive the most relevant chunks based on embedded metadata. Nemotron's embeddings map underlying concepts beyond keywords, helping users like lawyers surface related themes across different wording. The feature rolls out early next year for Enterprise Advance customers and combines five technologies into a single API.
