Infrastructure·
Box Integrates Nvidia Nemotron Embeddings for Agent‑First Content Retrieval
Box has embedded Nvidia's open‑weight Nemotron models into its platform, enabling AI agents to retrieve relevant document chunks up to four times faster while cutting token usage by 90%.
At its recent developer conference, Box unveiled Box Content Retrieval, a new feature that bakes Nvidia's Nemotron embedding models directly into the core of its content platform.
The integration lets AI agents pull the most relevant document segments instead of scanning entire files. Nvidia reports up to four‑times faster query responses, a 20% drop in error rates, and a 90% reduction in token consumption.
Box CEO Aaron Levie highlighted that Nemotron goes beyond keyword matching, mapping underlying concepts so agents can surface related legal clauses even when the wording differs.
Why it matters for GPU/AI infrastructure
- Open‑weight retrieval models like Nemotron cut the compute needed per query, easing GPU load.
- Lower token usage translates to cheaper inference and higher throughput on GPU clouds such as AiGpu.
- Unified API bundles metadata, permissions, and embedding lookup, simplifying deployment for enterprise teams.
The service will become available to Enterprise Advance customers early next year, giving organizations a ready‑made retrieval layer that leverages existing GPU resources.
- aigpu
- ai gpu
- ai gpu cloud
- aigpu dubai
- nvidia
- nemotron
- box
- embedding
By AiGpu Editorial · Editorial rewrite based on public reporting (TechCrunch AI)
← All articles