Box unveiled Box Content Retrieval at BoxWorks 2026, a new search solution that integrates Nvidia’s Nemotron embedding models. The tool is designed to enhance the accuracy of enterprise search capabilities across Box’s platform.
The company claims the new system improves search accuracy by 20 percent and reduces token consumption by up to 90 percent. Box will also integrate Nvidia’s Nemotron 3 Ultra large language model into its Box AI service.
The new features are available immediately to Enterprise Advanced and Shield customers. Box is also introducing a new orchestration tool called Box Architect for building agentic workflows. This tool helps teams design automated processes.
