NVIDIA Releases Expanded Agent Toolkit: Empowering Enterprise Agents with Data-Driven Customization

Summary:NVIDIA has launched an expanded version of its Agent Toolkit, empowering enterprises with enhanced capabilities for customizing intelligent agents. The toolkit integrates the NVIDIA Agent Knowledge Foundation, NVIDIA Kumo series models, NVIDIA OpenShell secure runtime, NVIDIA NemoClaw blueprints, and a suite of tools and skills. It enables enterprises to transform company data into actionable intelligence for agents while providing control over their operations and ensuring security.

---

NVIDIA Agent Toolkit Expansion: Key Features and Innovations

NVIDIA has launched an enhanced version of its Agent Toolkit, offering enterprises a more comprehensive suite of tools for developing and deploying intelligent agents. The key components and technical highlights of the toolkit include:

1. NVIDIA Agent Knowledge Foundation

   • This component integrates NVIDIA CUDA-X libraries, including cuDF, cuGraph, cuML, cuOpt, and cuVS, as well as the new Auto-Ontology library.
   • The Auto-Ontology library translates raw company data into familiar business concepts (e.g., customers, products, orders) and maps their relationships, enabling agents to better understand enterprise data.

2. NVIDIA Kumo Series Models

   • Includes Nemotron Retriever and Kumo Structured models (e.g., Kumo Tabular, Kumo Relational, Kumo Time Series, Kumo Forecast).
   • These models help agents extract relevant information from company data to answer questions and make predictions.

3. NVIDIA OpenShell

   • Provides a secure runtime environment, allowing enterprises to control what agents can access and what actions they can perform.

4. NVIDIA NemoClaw

   • Offers reference blueprint designs for building secure, specialized agents.
   • The latest NemoClaw blueprint focuses on agent optimization, providing a step-by-step starting point for improving agent intelligence over time.

5. Tools and Skills

   • Includes tools like NeMo Relay and NeMo Compass, supporting efficient agent operation across different scenarios.

Enterprise Adoption and Partnerships

The NVIDIA Agent Knowledge Foundation is already being utilized by various enterprises:

• Palantir: NVIDIA's supply chain organization uses Agent Knowledge Foundation through Palantir Foundry, Ontology, and AIP to deploy specialized agents, create actionable intelligence from proprietary data, predict outcomes, and optimize decision-making.
• Databricks: Pairs NVIDIA Kumo Anomaly with Databricks Genie Ontology to power context-aware agents.
• Snowflake: Enables customers to use the NVIDIA Kumo Engine and Kumo Relational models in their Snowflake Notebook environments to unlock more predictive capabilities.
• Amazon OpenSearch Service: Uses Nemotron Retriever models to improve search relevance, reduce search steps, and maintain low latency.
• SAP: Its fine-tuned reranker runs nearly 3x faster.
• ServiceNow: Improves reranking quality while cutting latency by nearly a quarter.
• Zoom: Uses Nemotron Retriever models to improve document understanding, search relevance, and context retrieval for Zoom AI Services.

Additionally, companies like AWS, ClickHouse, Cloudera, Cohesity, Dassault Systèmes, Dell, Elastic, Exa, IBM, NetApp, Oracle, Perplexity, Starburst, and VAST Data are also building on the Agent Knowledge Foundation.

Developer Recommendations and Deployment Advice

For developers looking to leverage the NVIDIA Agent Toolkit, here are some recommendations:

1. Data Preparation and Governance: Ensure the quality and governance of company data so that agents can effectively use it.
2. Model Selection and Fine-Tuning: Choose the appropriate Kumo model based on the specific application scenario and perform necessary fine-tuning to enhance performance.
3. Security and Privacy: Utilize the security features provided by OpenShell to ensure agents operate in a controlled environment and protect sensitive data.
4. Continuous Optimization and Monitoring: Use NemoClaw blueprints and tools to continuously optimize agent performance and monitor their behavior to ensure they meet expectations.

Conclusion

The expanded NVIDIA Agent Toolkit provides enterprises with a powerful way to transform company data into actionable intelligence for agents and enables the secure and controlled deployment of customized agents. This release marks a significant step forward for enterprises in the AI application domain, laying the foundation for the widespread use of intelligent agents in enterprise environments.

---

Source: NVIDIA AI News (2026-10-21)

— END —

Tags: #NVIDIA Follow #Intelligent Agents Follow #Enterprise AI Follow #AI Toolkit Follow #Data-Driven Follow