NVIDIA has announced the launch of NIM Agent Blueprints, a suite designed to accelerate the development of generative AI agents and applications by providing a catalog of workflows and software resources. Enterprise users can now download these blueprints for free, making it easy to quickly customize agents that suit their specific business needs.
These blueprints cover classic application scenarios such as customer service avatars, generative virtual screening for drug discovery, and multimodal PDF processing for data extraction from documents. Developers can leverage their own datasets to rapidly build and deploy these agents, with NVIDIA planning to add new application resources on a monthly basis.
In this era where enterprises are bullish on generative AI, NVIDIA's move is particularly significant. According to McKinsey, the deployment of generative AI by enterprises could annually generate $2.6 trillion to $4.4 trillion in value, covering over 60 application scenarios.
So, what can the NIM Agent Blueprints bring to enterprises? While many companies are already using generative AI for tasks like content generation and summarization, many teams aspire to go beyond these standard applications and create more complex ones using multiple AI agents and proprietary data. Although custom agents are considered the "second wave" of generative AI, the process of building and deploying them remains quite complex, with many organizations facing difficulties that slow progress and increase costs.
Through the NIM Agent Blueprints, NVIDIA provides teams with all the tools needed to accelerate workflows, including example applications built with NIM, NeMo, and partner microservices, reference code, customization documentation, and Helm charts for deployment. With pre-trained AI workflows in the blueprints, developers can easily initiate complex development processes and deploy their applications in accelerated data center and cloud environments. As users interact with these applications, the blueprints can also be continuously optimized, forming a continuous learning loop that improves application performance.
Currently, NVIDIA offers three main blueprints: digital human for customer service, generative virtual screening for accelerated drug discovery, and multimodal PDF data extraction for enterprise documents. The customer service blueprint allows enterprises to build 3D virtual customer service agents using NVIDIA's ACE, Omniverse RTX, Audio2Face, and Llama3.1NIM microservices. The drug discovery and data extraction blueprints combine Nvidia NeMo Retriever with NIM microservices including AlphaFold2, MolMIM, and DiffDock.
NVIDIA also collaborates with multiple technology solution providers such as Deloitte, Accenture, and World Power Technology, helping enterprises to more easily access and deploy these blueprints. Additionally, for enterprises wishing to independently deploy blueprints, NVIDIA provides full-stack accelerated infrastructure support, including NVIDIA-certified systems from Cisco, Dell, HP, and Lenovo, as well as accelerated cloud instances from Amazon AWS, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure.
Official Blog: https://blogs.nvidia.com/blog/nim-agent-blueprints/
Key Points:
🌟 NVIDIA introduces NIM Agent Blueprints, offering resources for quickly building generative AI applications.
💼 Three main blueprints cover customer service, drug discovery, and document data extraction.
🔗 NVIDIA collaborates with multiple technology solution providers to streamline the process of enterprise access and deployment of blueprints.