Nvidia releases NeMo tools for enterprise AI agent creation

Nvidia‘s new NeMo microservices suite represents a significant advancement in enterprise AI development, allowing companies to build custom AI agents that function as “digital employees” for various business tasks. By streamlining the complex processes of training, evaluating, and optimizing AI models, Nvidia is positioning these tools as productivity enhancers for the billion-plus knowledge workers across industries, with early implementations already showing measurable performance improvements in real-world applications.

The big picture: Nvidia has released NeMo microservices for general availability, offering enterprises comprehensive tools to develop, customize, and repeatedly optimize AI agents for specific business functions.

• The software is designed to create AI agents that act as “digital employees” capable of handling specialized tasks like customer service, billing inquiries, and software development.
• Early adopter Amdocs has already implemented NeMo-powered billing agents that achieved a 50% increase in first-call resolution rates for customer inquiries.

Key components: The NeMo suite consists of five integrated microservices that handle different aspects of the AI agent development lifecycle.

• Previously available components include Curator, which builds pipelines to clean and refine training data, and Retriever, which extracts usable elements like text and graphics from various data sources.
• Three new components complete the package: Customizer for fine-tuning models with new skills, Evaluator for benchmark testing improvements, and Guardrails for implementing runtime safety and security measures.

How it works: Nvidia designed NeMo as a “flywheel” system where models repeatedly cycle through the microservices to gain new capabilities and improve performance.

• The NeMo tools integrate with Nvidia’s Inference Microservices (NIM) infrastructure software for seamless deployment of the finished agents.
• According to Nvidia’s Joey Conway, the system dramatically simplifies tasks that would otherwise require developers to work directly with complex Python code and AI frameworks.

Why this matters: By packaging AI agent development into accessible microservices, Nvidia is lowering the technical barriers for enterprises wanting to implement advanced AI solutions across their operations.

• The approach reflects Nvidia’s vision of AI agents augmenting the global knowledge workforce, potentially transforming how companies handle specialized tasks and customer interactions.
• The measurable productivity gains reported by early adopters suggest these AI agents could provide significant operational improvements for businesses investing in the technology.