# Box adds Nvidia Nemotron to new Box Content Retrieval feature for agents

published: 2026-11-05T15:05:45.572Z
topic: search
source: https://techcrunch.com/2026/11/05/box-taps-nvidias-nemotron-for-new-agent-first-content-retrieval-system/
url: https://aiagentnews.io/stories/box-adds-nvidia-nemotron-to-new-box-content-retrieval-feature-for-agents

Box showed Box Content Retrieval at its developer conference, a feature that builds Nvidia's Nemotron embedding system into its core product. Box says agents using it will get faster, more accurate and cheaper answers to queries, and it rolls out to Enterprise Advance customers early next year.

Box used its developer conference to show Box Content Retrieval, a feature that builds Nvidia's Nemotron embedding system into the core of the Box product. Agents answer queries using the embedded metadata that Nemotron creates, and Box says this makes queries faster, more accurate and cheaper to run. Nvidia's VP of generative AI software, Kari Briski, told TechCrunch she expects answers four times faster, with 20% fewer errors and 90% fewer tokens used.

The feature implements Nvidia's Nemotron Embed project, which offers open-weight embedding models for automated retrieval. Box CEO Aaron Levie said that instead of having an agent read each document, the system can return the chunks most relevant to a user's query or task. Box expects the feature to help lawyers find relevant clauses and precedents, since the embeddings map concepts rather than only keywords.

Box Content Retrieval will be available to all Enterprise Advance customers, with rollout starting early next year. The Nemotron models are freely available, but the system depends heavily on Box's existing metadata and permission controls, which would make it hard to recreate from scratch. Levie said the company is packing five different technologies into one API.

## Why it matters if you build agents

Developers building agents on Box may get retrieval that returns relevant passages instead of whole documents, which could lower token costs if the reported figures hold. Teams building similar pipelines should note that the gains rely on Box's metadata and permission layer, not only on the open Nemotron models. The feature is not expected to reach Enterprise Advance customers until early next year.