Box 采用 Nemotron 构建智能体优先检索系统

在周四的开发者大会上,Box 展示了一项新功能,该功能将把 NVIDIA 的 Nemotron 嵌入系统构建进其产品核心。
这项新功能名为 Box Content Retrieval,它让智能体能够依托 Nemotron 创建的嵌入元数据,更高效地回答查询。
Box 表示,对用户而言,其结果是查询更快、更准确、成本更低。NVIDIA 生成式 AI 软件副总裁 Kari Briski 告诉 TechCrunch,她预计查询的响应速度将提升 4 倍,错误减少 20%,Token 使用量减少 90%。
该系统将面向 Box 所有 Enterprise Advance 客户提供,于明年年初推出。
“这极大地提升了我们搜索技术对智能体的相关性,”Box CEO Aaron Levie 表示。“智能体不必再逐份通读每个文档,我们可以直接返回文档中与用户查询或智能体所执行任务最相关的片段。”
该功能是 NVIDIA Nemotron Embed 项目的一项落地实现,该项目提供开放权重嵌入模型以改进自动化检索系统。这类模型利用 LLM 对数据进行预处理,以便智能体更好地使用,但大型实验室往往忽视了它们。这一领域一直是 NVIDIA 开源模型项目的重点,其 7 月发布的最新 Nemotron 在衡量检索质量的 RTEB 基准测试中位居榜首。
Box 表示,该系统与主动型智能体配合使用时最为有用,并以律师为例:在新系统下,他们表面上能够更轻松地找到相关条款和先例。由于 Nemotron 的嵌入向量不止于简单关键词,而是能映射底层概念,因此即使使用了不同的措辞或语言,搜索工具也能呈现相关的法律主题。
虽然 Nemotron 模型可免费获取,但该系统也严重依赖 Box 既有的元数据和权限控制,这使得从头复刻它颇具挑战。“我们大概是把五种不同的技术打包进了一个简单的 API 里,”Levie 对 TechCrunch 表示。
当你通过我们文章中的链接购买时,我们可能会赚取少量佣金。这不会影响我们编辑的独立性。
Russell Brandom 自 2012 年以来一直报道科技行业,重点关注平台政策与新兴技术。他此前曾任职于 The Verge 和 Rest of World,并为 Wired、The Awl 和《麻省理工科技评论》撰稿。可通过 [email protected] 或 Signal 412-401-5489 联系他。
正在加载下一篇文章
译文由英文原文机器翻译,术语按行业惯例统一,可在文末展开原文核对。
查看英文原文(Box taps Nvidia’s Nemotron for new agent-first content retrieval system)
Box taps Nvidia’s Nemotron for new agent-first content retrieval system | TechCrunch
Image Credits:Steve Jennings/Getty Images for TechCrunch
Box taps Nvidia’s Nemotron for new agent-first content retrieval system
6:00 AM PST · November 5, 2026
At its developer conference on Thursday, Box showed off a new feature that will build Nvidia’s Nemotron embedding system into the core of its product.
Called Box Content Retrieval, the new feature allows agents to answer queries far more effectively by relying on the embedded metadata created by Nemotron.
Box says the upshot for users is faster, more accurate, and less expensive queries. Nvidia’s VP of generative AI software, Kari Briski, told TechCrunch that she expects queries to be answered four times faster, with 20% fewer errors and 90% fewer tokens used.
The system will be available to all of Box’s Enterprise Advance customers, rolling out early next year.
“This dramatically boosts both the relevance of our search technology for agents,” Box CEO Aaron Levie said. “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.”
The feature is an implementation of Nvidia’s Nemotron Embed project, which offers open-weight embedding models to improve automated retrieval systems. Major labs have often overlooked such models, which use LLMs to pre-process data so it can be better used by agents. The sector has been a focus for Nvidia’s open source model project, with its recent Nemotron release in July topping the RTEB benchmark, which measures retrieval quality.
The system will be most useful when paired with active agents, Box says, giving the example of lawyers, who will ostensibly be able to find relevant clauses and precedents more easily under the new system. Because Nemotron’s embeddings go beyond simple keywords to map underlying concepts, a search tool can surface related legal themes even when different wording or language is used.
While the Nemotron models are freely available, the system also relies heavily on Box’s pre-existing metadata and permission controls, which will make it challenging to recreate from scratch. “We’re probably packing in five different technologies into one simple API,” Levie told TechCrunch.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Russell Brandom has been covering the tech industry since 2012, with a focus on platform policy and emerging technologies. He previously worked at The Verge and Rest of World, and has written for Wired, The Awl and MIT’s Technology Review. He can be reached at [email protected] or on Signal at 412-401-5489.
Loading the next article
Error loading the next article
原文为英文,正文由自动抓取并保留原文。
原文发布于 TechCrunch AI(Box taps Nvidia’s Nemotron for new agent-first content retrieval system),标题、摘要与正文由机器翻译,术语按行业惯例统一,未经人工逐句校订。