NVIDIA Unveils Secure AI Agents with Enterprise RAG Blueprints

In a strategic move to bolster the capabilities of generative AI, NVIDIA has launched the AI-Q Research Assistant and Enterprise RAG Blueprints. These advanced tools are designed to create AI agents that are not only accurate and reliable but also tailored to the specific data needs of businesses, according to NVIDIA.

Enhancing AI with Retrieval-Augmented Generation

The core of this innovation lies in the use of retrieval-augmented generation (RAG) and NVIDIA Nemotron reasoning AI models. These technologies automate document comprehension, extract valuable insights, and facilitate the generation of high-value reports from extensive datasets. This approach is particularly beneficial for organizations seeking to leverage AI for more informed decision-making and data analysis.

Infrastructure and Deployment on AWS

Deploying these AI agents requires a robust infrastructure that is not only secure but also scalable. NVIDIA’s blueprints are optimized for deployment on Amazon Web Services (AWS), utilizing the Amazon Elastic Kubernetes Service (EKS) for orchestration. This setup ensures that the AI agents can operate efficiently while maximizing cost-effectiveness. The infrastructure also incorporates Amazon OpenSearch Serverless for vector database management, Amazon Simple Storage Service (S3) for data storage, and Karpenter for dynamic GPU scaling.

Blueprint Components and Architecture

The AI-Q Research Assistant builds on the Enterprise RAG Blueprint, which serves as the foundational component. These blueprints consist of NVIDIA NIM microservices, optimized for high-throughput, low-latency AI model performance on GPUs. Key components include large language models (LLM) such as Llama-3.3-Nemotron-Super-49B-v1.5, and NeMo Retriever models for advanced data retrieval capabilities.

The architecture is designed to deploy these components as pods within a Kubernetes cluster. This setup allows for dynamic provisioning of GPU instances, optimizing both performance and cost.

AI-Q Research Assistant Workflow

The AI-Q blueprint enhances the RAG foundation by integrating a sophisticated agentic workflow. This workflow includes planning, refining, and reflecting stages, allowing the AI agents to generate comprehensive reports based on real-time data and existing enterprise knowledge. The integration of web search capabilities via the Tavily API ensures that the reports are based on the latest available information.

Conclusion

NVIDIA’s introduction of the AI-Q Research Assistant and Enterprise RAG Blueprints marks a significant advancement in the development of data-driven AI agents. By leveraging AWS infrastructure, these solutions provide a secure and scalable environment for deploying AI applications that can transform enterprise data into actionable insights. Organizations can now deploy these blueprints on Amazon EKS to enhance their data processing and decision-making capabilities.

Image source: Shutterstock