Box Taps Nvidia Nemotron for Agent-First Content Retrieval

Box used its developer conference on Nov 5 to unveil Content Retrieval, a new system that plugs Nvidia's Nemotron embeddings into Box so enterprise AI agents can pull the most relevant chunks of a document instead of ingesting the whole file. TechCrunch reported the integration, with the performance numbers attributed to Nvidia rather than an independent benchmark.

The headline claim, from Nvidia VP of generative AI software Kari Briski, is that the setup lets "queries to be answered four times faster, with 20% fewer errors and 90% fewer tokens used." Box CEO Aaron Levie framed the point in plainer terms: "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." He also described the product as a bundle, saying Box is "probably packing in five different technologies into one simple API."

Box says Content Retrieval will be available to all Enterprise Advance customers, rolling out early next year. It lands in a busy stretch of enterprise-agent news on our radar, alongside this week's Sierra/Anon acquisition — part of a run of enterprise AI coverage where retrieval and agent action are increasingly sold as one layer.