Box builds Nvidia Nemotron into new agent-focused content retrieval feature
At its developer conference on 5 November, Box unveiled a retrieval system built on Nvidia's Nemotron models, which it says is faster, more accurate and cheaper for AI agents. It also announced a workflow builder and new threat protection.
1 / 2
The story, neutrally told
Centre · 2At its developer conference on Thursday 5 November, Box unveiled a retrieval feature that builds Nvidia's Nemotron embedding system into the core of its product. TechCrunchN “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.” Read at TechCrunch ↗ CNAN “Content management firm Box expanded its partnership with Nvidia on Thursday” Read at CNA ↗ Centre · 2TechCrunch calls the feature Box Content Retrieval, while CNA calls it Box Context Retrieval. It lets agents answer queries by relying on metadata embedded by Nemotron. TechCrunchN “Called Box Content Retrieval, the new feature allows agents to answer queries far more effectively by relying on the embedded metadata created by Nemotron.” Read at TechCrunch ↗ CNAN “Box Context Retrieval, a new search engine built on Nvidia's Nemotron models” Read at CNA ↗ Centre · 2Nvidia's VP of generative AI software, Kari Briski, told TechCrunch she expects queries to be answered four times faster, with 20% fewer errors and 90% fewer tokens used. CNA reports Box's own figures as 20 per cent better search accuracy, 90 per cent lower token use and fourfold faster question-answering. TechCrunchN “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.” Read at TechCrunch ↗ CNAN “Box said the engine could improve search accuracy by 20 per cent, cut token usage by 90 per cent and speed up question-answering fourfold” Read at CNA ↗
Centre · 1Box CEO Aaron Levie said that instead of an agent having to read through each document, Box can send back the chunks most relevant to the user's query or the agent's task. TechCrunchN “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.” Read at TechCrunch ↗ Centre · 2The system is due to reach Box's Enterprise Advanced customers early next year. TechCrunchN “The system will be available to all of Box’s Enterprise Advance customers, rolling out early next year.” Read at TechCrunch ↗ CNAN “will be generally available to Enterprise Advanced customers early next year, the company said” Read at CNA ↗ Centre · 1TechCrunch describes the feature as an implementation of Nvidia's Nemotron Embed project, which offers open-weight embedding models. It says Nvidia's Nemotron release in July topped the RTEB retrieval-quality benchmark. TechCrunchN “The feature is an implementation of Nvidia’s Nemotron Embed project, which offers open-weight embedding models to improve automated retrieval systems.”“its recent Nemotron release in July topping the RTEB benchmark, which measures retrieval quality” Read at TechCrunch ↗
Centre · 1Box gave lawyers as an example of likely users, who it says should find relevant clauses and precedents more easily. Because Nemotron's embeddings map underlying concepts rather than keywords, a search can surface related themes even when the wording or language differs. TechCrunchN “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.” Read at TechCrunch ↗ Centre · 1TechCrunch notes the Nemotron models are freely available but the system relies heavily on Box's existing metadata and permission controls, which would make it hard to recreate from scratch. Levie said it packs "five different technologies into one simple API". TechCrunchN “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.” Read at TechCrunch ↗ Centre · 1Box also introduced Box Architect, which lets teams build automated workflows by describing a process in plain language, with a beta rollout to Enterprise Advanced customers set for the fourth quarter. It announced threat protection against bulk downloads, account takeovers and mass deletion, due early next year for Enterprise Advanced and Shield customers. CNAN “The company also introduced Box Architect, which lets teams build automated workflows by describing a process in plain language and is set to begin rolling out in beta to Enterprise Advanced customers in the fourth quarter”“New threat protection against bulk downloads, account takeovers and mass deletion will be available to Enterprise Advanced and Shield customers early next year, Box added” Read at CNA ↗
Every sentence links to the reporting it rests on. The pill in front of each says where its sources sit: Left, Centre or Right when one side supplies at least half of them, Mixed when they are evenly split. The number is how many outlets it cites.
Left0 outlets
No left outlet in our sources has covered this story yet.
Centre2 outlets
- Framing
- Both outlets treat it as a product announcement. TechCrunch focuses on the Nemotron integration and what it means for agents. CNA gives a short roundup that also covers Box's other announcements.
- Emphasis
- TechCrunch stresses how the technology works, the lawyer use case and the difficulty of recreating Box's metadata and permissions. CNA stresses availability dates and the additional Box Architect and threat-protection products.
- Leaves out or plays down
- CNA leaves out the technical background and quotes from Levie and Briski. TechCrunch does not mention Box Architect or the threat-protection features.
- Charged language
- “supposedly”“ostensibly”
- For example
-
“The system can supposedly answer queries four times faster, with 20% fewer errors and 90% fewer tokens used.” — TechCrunch
“Box expands Nvidia partnership, unveils AI workflow builder” — CNA
Right0 outlets
No right outlet in our sources has covered this story yet.
What every side reports
- Box announced the retrieval system on 5 November, built on Nvidia's Nemotron models.
- It is due for Enterprise Advanced customers early next year.
- The headline performance figures are fourfold faster answers, 20% better accuracy or fewer errors, and 90% fewer tokens.
Where accounts differ
-
Name of the feature and who the performance figures come from
- Centre
- TechCrunch calls it Box Content Retrieval and attributes the figures to Nvidia's Kari Briski as expectations. CNA calls it Box Context Retrieval and attributes the figures to Box, saying the engine 'could' achieve them. Both are projections rather than independently tested results.
Box organisation
Box presents the feature as a major improvement in search relevance for agents, packing several technologies into one API on top of its metadata and permission controls. It is also expanding its Nvidia partnership and adding a workflow builder and security features.
““This dramatically boosts both the relevance of our search technology for agents,” Box CEO Aaron Levie said.” — TechCrunch
“Content management firm Box expanded its partnership with Nvidia on Thursday” — CNA
Nvidia organisation
Nvidia, through VP Kari Briski, expects faster, more accurate and cheaper queries from its Nemotron embedding models. TechCrunch notes its July Nemotron release topped the RTEB benchmark.
“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.” — TechCrunch
North America1 outlet
TechCrunch gives a detailed, product-focused account that includes sourced quotes and technical context.
“Called Box Content Retrieval, the new feature allows agents to answer queries far more effectively” — TechCrunch
TechCrunch
East and Southeast Asia1 outlet
CNA runs a brief business roundup of the partnership expansion and Box's other announcements.
“Box expands Nvidia partnership, unveils AI workflow builder” — CNA
CNA
Left0 articles
No coverage yet.
Centre2 articles
-
Box taps Nvidia’s Nemotron for new agent-first content retrieval system
Neutral Explains the Nemotron integration and its use for agents, with some scepticism in 'supposedly' and 'ostensibly' about the performance and legal use-case claims.
-
CNAN ·
Box expands Nvidia partnership, unveils AI workflow builder
Neutral Short bulleted roundup of Box's announcements, attributing the performance figures to the company.
Right0 articles
No coverage yet.
- 5 Nov 14:00 First TechCrunchN Box taps Nvidia’s Nemotron for new agent-first content retrieval system
- 5 Nov 14:06 +6m CNAN Box expands Nvidia partnership, unveils AI workflow builder
Times are when each article was published, or when we first saw it if the outlet gave no time.
Ask about this story
Loading…