Nvidia Unveils New Microservices to Enhance Safety and Control of AI Agents

Nvidia Introduces New Microservices for AI Agent Safety

Nvidia has announced the release of new Nvidia Inference Microservices (NIM) as part of its NeMo Guardrails collection, aimed at enhancing the safety, security, and control of AI agents. This development comes as enterprises increasingly seek to adopt AI technologies while addressing critical concerns surrounding trust, safety, and compliance 1

1

2

Key Features of the New Microservices

The newly introduced NIM microservices focus on three primary areas:

1. Content Safety: Trained using the Aegis Content Safety Dataset, this microservice helps moderate content to prevent harmful or inappropriate outputs 1

   1

   2

   2

   .
2. Topic Control: Designed to keep AI agents focused on their intended purpose, preventing conversations from straying off-topic 1

   1

   3

   3

   .
3. Jailbreak Protection: This feature safeguards against attempts to bypass system restrictions, enhancing overall security 1

   1

   4

   4

   .

These lightweight, specialized models are designed to work efficiently in various environments, including resource-constrained settings, making them ideal for scaling AI applications across industries such as healthcare, automotive, and manufacturing 2

2

Benefits and Performance Improvements

Nvidia claims that the new NIM microservices offer significant advantages:

• 50% better protection with only a half-second increase in latency 4

  4

  .
• Ability to stack multiple guardrails with minimal additional response time 1

  1

  .
• Improved scalability for AI applications in diverse industrial settings 2

  2

  .

Industry Adoption and Use Cases

Several industry leaders are already leveraging NeMo Guardrails to enhance their AI applications:

• Amdocs: Implementing the technology to improve AI-driven customer interactions in the communications and media sector 2

  2

  .
• Cerence AI: Utilizing NeMo Guardrails to ensure safe and context-aware interactions in automotive AI assistants 2

  2

  .
• Lowe's: Employing the technology to empower store associates with enhanced product knowledge while maintaining safety and reliability 2

  2

  .

Related Stories

Nvidia Launches NeMo Microservices to Boost Enterprise AI Agent Development

Nvidia Releases Open-Source AI Safety Platform After Rogue Agent Incidents Expose Security Gaps

NVIDIA Unveils Agentic AI Blueprints and Models to Revolutionize Enterprise Automation

Nvidia launches Nemotron 3 Super with 5x throughput boost to power enterprise AI agents

Addressing Enterprise Concerns

The introduction of these microservices addresses key barriers to enterprise AI adoption, including:

• Security and data privacy concerns 4

  4

  .
• Governance requirements 4

  4

  .
• Performance and latency issues 1

  1

  4

  4

  .

By providing more granular controls and optimized performance, Nvidia aims to make it easier for organizations to deploy AI agents confidently 3

3

4

Availability and Licensing

The new NeMo Guardrails NIMs are available under the Nvidia AI Enterprise license, priced at $4,500 per GPU per year. Developers can also try them out for free under an open-source license or on build.nvidia.com 4

4

As AI continues to transform various industries, Nvidia's latest offering represents a significant step towards creating more secure, trustworthy, and efficient AI agents. By addressing critical concerns surrounding AI safety and control, these microservices may help accelerate the adoption of AI technologies across enterprises.