NVIDIA Expands Agent Toolkit to Enhance Enterprise Data Grounding

NVIDIA has significantly expanded its open-source Agent Toolkit, introducing new components such as the Agent Knowledge Foundation and the Kumo family of models. This update is designed to help enterprises deploy AI agents that are grounded in proprietary company data while maintaining strict governance and security controls. By providing tools that allow agents to access and process internal information securely, NVIDIA aims to bridge the gap between general-purpose foundation models and the specific, high-stakes requirements of corporate environments. The toolkit addresses the critical need for 'grounding'—ensuring that AI agents provide accurate, context-aware responses based on verified internal documentation rather than hallucinated data. This development marks a shift toward more reliable, enterprise-native AI platforms that prioritize data integrity and operational control.