The GPU-powered software suite enables businesses to drive growth and security while cutting costs, thanks to AI.
Late last week NVIDIA announced plans to launch AI Enterprise 3.0, a GPU-powered software suite for business-focused AI workflows including virtual assistants, audio transcription, digital fingerprinting for cybersecurity and support for more than 50 of NVIDIA’s AI frameworks, pretrained models, Helm charts, Jupyter notebooks and other tools.
NVIDIA AI Enterprise 3.0 is geared to accelerate enterprise data science initiatives and streamline AI model deployment to enable rapid insights from data analytics. The suite’s various micro-services can run as stand-alone Kubernetes containers or be combined with other services to create production-ready applications with accuracy and performance, according to a NVIDIA blog post announcing the new cloud-native software suite.
NVIDIA says it will include a license of AI Enterprise 3.0 for server hardware equipped with its H100 PCIe GPUs from partners including Lenovo, HPE, Dell and Supermicro.
How Enterprises Can Leverage AI Today
Enterprises are showing interest in AI tools for customer-facing applications to help offset labor shortages and supply chain disruptions, cut costs, and improve operational efficiency. However, hardware and software integrations are prerequisites necessary for effective deployment.
According to analyst McKinsey’s State of AI in 2022 report, AI adoption has more than doubled since 2017 and 63 percent of businesses expect investment in AI to increase over the next three years. Even though the share of organizations adopting AI has plateaued at between 50 and 60 percent in recent years, those that have adopted the technology have increased the scope of their AI capabilities.
AI Enterprise 3.0 includes unencrypted pretrained models and source code from NVIDIA’s TAO Toolkit, a low-code AI development solution for customized, production-ready AI models for speech and computer vision applications that can be used in health care, smart cities, retail, and other uses. AI Enterprise 3.0 also provides a wide range of other AI frameworks and infrastructure options, including NVIDIA Clara Parabricks, which supports faster genomic analysis and MONAI for medical imaging.
Additionally, AI Enterprise 3.0 provides Riva for GPU-accelerated speech AI software development; Morpheus for cybersecurity developers to create optimized AI pipelines to filter, process and classify large datasets; Metropolis for video analytics; DeepStream for vision AI; and Merlin for building high-performing recommender systems at scale.
Mounting cybersecurity attacks have put security at the forefront of enterprise priorities. NVIDIA AI Enterprise 3.0 includes a digital fingerprinting AI workflow that enhances security by helping enterprises fingerprint all users, services, accounts and machines across networks and detect potential threats. The workflow uses unsupervised learning to provide alerts and actionable information while reducing detection time from weeks to minutes, giving cybersecurity professionals stronger opportunity to take action against threats.
The NVIDIA audio transcription AI workflow uses automation speech recognition technology to provide transcripts in English, Spanish, Mandarin, Hindi, Russian, Korean, German, French, and Portuguese, with Japanese, Arabic and Italian languages coming soon. It relies on fully customizable GPU-optimized models for better understanding, contextual insights and sentiment analysis with real-time accuracy. The transcripts can be used to improve product development and contact center agent training, among other uses.
Customer service automation tools are a key focus area for many enterprises, according to NVIDIA, whose virtual assistant AI workflow provides tools for enterprises to respond to customer needs 24/7 while human contact center agents are freed up to respond to more complex issues. It not only reduces costs but also allows agents to better understand customers when call audio quality is poor, and offers personalized and precise responses via natural-sounding artificial voices.
