Box has deepened its collaboration with Nvidia by introducing Box Content Retrieval, a new AI-powered engine that streamlines access to unstructured enterprise data, enhancing the contextual intelligence and security of autonomous AI agents across organizations.
- New retrieval engine boosts search accuracy by 20% and cuts token use by 90%
- AI agents gain secure, permission-based access to vast unstructured corporate data
- Nvidia’s open-weight Nemotron 3 Ultra model integrates for affordable, domain-specialized AI workflows
What happened
Box announced a significant enhancement to its AI capabilities with the introduction of Box Content Retrieval, a new engine developed in partnership with Nvidia. It uses Nvidia’s Nemotron Retriever embedding models to convert messy, unstructured enterprise data into an organized knowledge base accessible by AI agents. This enables autonomous agents to better understand and reason over diverse data types such as documents, images, audio, and video.
Alongside the retrieval engine, Box revealed an orchestration tool designed for managing complex multi-step AI workflows and introduced threat detection technologies to protect organizations from unauthorized data access by rogue agents. The initiative aims to tackle the challenges enterprises face when deploying autonomous AI by providing both context richness and data security.
Why it matters
AI-driven autonomous agents rely heavily on context to perform tasks accurately and efficiently. However, much of the necessary context resides in unstructured and sensitive corporate data, which has been difficult to safely expose to AI systems. Without addressing this, enterprises risk errors and data breaches, impeding AI adoption. Box’s solution addresses this bottleneck by ensuring AI agents only access content consistent with user permissions, thereby preserving data confidentiality.
Additionally, Box’s use of Nvidia’s open-weight AI models such as Nemotron 3 Ultra offers cost-effective, domain-optimized language processing. This development allows specialized teams—like legal and financial analysts—to leverage AI insights precisely tailored to their workflows, increasing operational efficiency without the price premiums often associated with proprietary AI models.
What to watch next
Enterprises will be monitoring the real-world impact of Box’s AI enhancements on workflow automation, particularly how the improved contextual retrieval and secure access translate into productivity gains and stronger data governance. Early adoption success stories may set new standards for trust and functionality in autonomous enterprise AI deployments.
The partnership’s progression will also depend on how well Box integrates Nvidia’s open-weight AI models with leading cloud platforms and adapts to evolving enterprise security requirements. Future releases may expand the scope of AI orchestration and threat detection, further refining the balance between powerful automation and rigorous data protection.