{
  "id": 18317758,
  "title": "Box taps Nvidia’s Nemotron for new agent-first content retrieval system",
  "url": "https://urgent.news/2026/11/05/box-taps-nvidias-nemotron-for-new-agent-first-content-retrieval-system",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-11-05T14:00:00.000Z",
  "source": {
    "name": "TechCrunch",
    "slug": "techcrunch",
    "url": "https://techcrunch.com/2026/11/05/box-taps-nvidias-nemotron-for-new-agent-first-content-retrieval-system/"
  },
  "original_language": "en",
  "account": "At its developer conference, Box unveiled a new feature called Box Content Retrieval, which integrates Nvidia’s Nemotron embedding system into its product. This innovation enables agents to provide more accurate and efficient answers to queries by utilizing embedded metadata generated by Nemotron.\n\nAccording to Box, users can anticipate faster, more precise, and less expensive queries. Nvidia’s Vice President of Generative AI Software, Kari Briski, expects query response times to improve by four times, with a 20% reduction in errors and a 90% decrease in token usage.\n\nBox will make this system available to all of its Enterprise Advance customers, with an early rollout anticipated in the coming year. Aaron Levie, Box’s CEO, emphasized that this enhancement significantly improves the relevance of their search technology for agents. Rather than agents manually reading through each document, the system can now retrieve and present the most pertinent information based on user queries or tasks.\n\nThis feature is built upon Nvidia’s Nemotron Embed project, which offers open-weight embedding models to enhance automated retrieval systems. The sector has largely overlooked such models, which employ large language models (LLMs) to preprocess data, making it more efficient for agents to utilize.\n\nNvidia has been proactively focusing on open source model projects, with its recent Nemotron release in July achieving top rankings on the RTEB benchmark, which assesses retrieval quality. The system is anticipated to be particularly beneficial when used in conjunction with active agents, such as lawyers. Under this new system, lawyers could more easily locate relevant clauses and precedents.\n\nThe incorporation of Nemotron’s embeddings, which extend beyond mere keywords to map underlying concepts, allows the search tool to uncover related legal themes even when different wording or language is employed. Although the Nemotron models are freely accessible, the system heavily relies on Box’s existing metadata and permission controls, which may pose challenges for independent replication. Levie stated that this integration essentially combines five different technologies into a single, user-friendly API.",
  "summary": "The system can supposedly answer queries four times faster, with 20% fewer errors and 90% fewer tokens used.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}