Box adds Nvidia Nemotron to speed up agent content retrieval
Box's new Content Retrieval feature uses Nvidia's Nemotron embeddings to make agent queries faster, more accurate and cheaper.
At its developer conference, Box announced Box Content Retrieval, a feature that integrates Nvidia's Nemotron embedding system into its platform. The feature lets AI agents answer user queries by pulling the most relevant document chunks rather than scanning entire files.
Nvidia's VP of generative AI software said the system can answer queries four times faster, cut errors by 20% and reduce token usage by 90%. The feature will roll out to Box Enterprise Advance customers early next year.
Why it matters to your business
For founder‑led businesses that rely on AI agents for customer support, sales or legal work, faster and more precise retrieval cuts response times and operational costs. Better relevance also means fewer manual checks and higher productivity.
The integration leverages Box's existing metadata and permission controls, meaning you get the benefit without rebuilding complex search infrastructure.
What to do on Monday
- Evaluate whether your team uses Box Enterprise Advance and map out use cases where agents handle document searches.
- Plan a pilot with Box's Content Retrieval for a high‑volume workflow, such as contract review or support tickets, to measure speed and error improvements.
- If the pilot shows gains, schedule a rollout and train agents on prompting techniques that exploit the new chunk‑level relevance.
Sources
Reported by techcrunch.com.